Camber-Car Design & Dynamics Document Optimized 40 Character Title

Camber-Car Design & Dynamics
Tristan French, E.I.T.

tristan.french@gmail.com
Alissa Roland, E.I.T.
alissaroland3@gmail.com
Subsystem Lead:
Maximilian Sluiter, E.I.T.
maximiliansluiter@gmail.com
Mechanical Engineering Department

California Polytechnic State University
San Luis Obispo
2013
Contents
List of Figures ...................................................................................................................... i
Abstract & Background ...................................................................................................... 1
Fundamental Tire & Vehicle Dynamics ............................................................................. 2
Suspension Basics ........................................................................................................... 2
Tire Basics ...................................................................................................................... 2
Slip Angle ....................................................................................................................... 4
Inclination ....................................................................................................................... 7
Combining Inclination & Slip Angle ............................................................................ 10
Round-Section Tires vs. Square-Shouldered Tires ....................................................... 16
Longitudinal Force ........................................................................................................ 20
Induced Drag ................................................................................................................. 22
Effects of Camber on Vehicle & Suspension Design ................................................... 22
Transient Response ....................................................................................................... 31
Conclusions Regarding Vehicle and Suspension Design ............................................. 32
Dynamic Camber .............................................................................................................. 36
Active vs. Passive ......................................................................................................... 36
Polynx ........................................................................................................................... 38
Mechanism Design........................................................................................................ 39
Design Direction Change .............................................................................................. 40
Final Design: Static Camber ............................................................................................. 41
Kinematics .................................................................................................................... 41
Component Design & Analysis .................................................................................... 52
Recommended Revisions .............................................................................................. 64
Design Verification ........................................................................................................... 65
Structural Testing .......................................................................................................... 65
Testing Procedure ......................................................................................................... 69
References ......................................................................................................................... 71
Appendix A: Simulink Modeling...................................................................................... 72
Appendix B: Vendors ....................................................................................................... 73
List of Figures
Figure 1: SAE Tire Axes................................................................................................................. 3
Figure 2: Tire Vertical Load, Lateral Force, and Contact Patch Deformation Distribution for Slip
Angle ................................................................................................................................... 5
Figure 3: Typical Load Sensitivity Behavior .................................................................................. 6
Figure 4: Shape of Deformed Contact Patch for Frictionless Road................................................ 8
Figure 5: Changes in Contact Patch Area and Shape due to Camber and Tread Blocks ................ 9
Figure 6: Load Sensitivity of Motorcycle Tires with Slip Angle but None for Inclination............ 9
Figure 7: Lateral Force of Motorcycle 120/70-17 Front Tire at Combined Slip and Inclination
Angles ............................................................................................................................... 10
Figure 8: Milliken MX-1 Front View ........................................................................................... 11
Figure 9: Milliken MX-1 Side View ............................................................................................. 11
Figure 10: Lateral Force for Combined Camber and Slip Angles at Low Load (291 pounds) for
5.00-16 Bias-Ply, Ribbed-Tread Goodyear Eagle Motorcycle Tires ................................ 12
Figure 11: Lateral Force from 5.00-16 Bias-Ply, Ribbed-Tread Goodyear Eagle Motorcycle Tires
at Combined Inclination and Slip Angles with Vertical Load of 492 pounds .................. 13
Figure 12: Simulation Comparing Effect of Lateral Weight Transfer on Total Lateral Force from
A Pair of Tires With and Without Negative Camber ........................................................ 15
Figure 13: Simulation of Lateral Force for Pair of Tires at -40 Degrees of Camber Showing More
Lateral Force than -20 Degrees ......................................................................................... 16
Figure 14: Load Sensitivity for Various Camber Angles at 7 Slip Angle .................................. 17
Figure 15: Peak Lateral Force vs. Inclination Angle .................................................................... 18
Figure 16: Tires Used to Generate Data for Figure 9 ................................................................... 18
Figure 17: Lateral Force Characteristics for Various Round-Shouldered Tires ........................... 19
Figure 18: Friction Circle from Sennas Lotus at 1987 Australian GP ..................................... 21
Figure 19: Ackermann Steering Geometry ................................................................................... 23
Figure 20: Steering Geometry Definitions .................................................................................... 25
Figure 21: Instant Center vs. True Center ..................................................................................... 28
Figure 22: Roll Center Construction ............................................................................................. 29
Figure 23: Vertical Jacking Force Due to Applied Lateral Force ................................................. 29
Figure 24: General Polynx Design ................................................................................................ 39
i
Figure 25: Steering Mechanism Shown on a Simplified (2 Pivot) Polynx Suspension................ 40

Figure 26: Front Suspension Pivot Point Front View Geometry .................................................. 42
Figure 27: Rear Suspension Pivot Point Front View Geometry ................................................... 42
Figure 28: Side View of Front Suspension Points ........................................................................ 43
Figure 29: Side View of Rear Suspension Points ......................................................................... 44
Figure 30: Front View of Front Upright Geometry ...................................................................... 44
Figure 31: Front View of Rear Upright Geometry ....................................................................... 45
Figure 32: Top View of Left Front Upright Lower Clevis Plate .................................................. 46
Figure 33: Top View of Left Front, High Camber Upright .......................................................... 46
Figure 34: Side View of Rear Upright Geometry ......................................................................... 47
Table 1: Link Lengths ................................................................................................................... 48
Figure 35: Ackermann of the Front Multilink Steering System ................................................... 49
Figure 36: Bump Steer for Multilink with Rack & Pinion Steering ............................................. 50
Figure 37: Final Suspension Design ............................................................................................. 53
Figure 38: Right Rear Suspension Corner .................................................................................... 53
Figure 39: Anti-Bumpsteer Steering Rack Clevises ..................................................................... 55
Figure 40: Front Pushrod Rotating Clevis and Bushing Area ...................................................... 57
Figure 41: Front Uprights and Links ........................................................................................... 57
Figure 42: Installed Rotating Clevis ............................................................................................. 58
Figure 43: Front (High Camber) Lower Control Arm Pivots, Steering Pivot, and Heim Joint
Spacers .............................................................................................................................. 58
Figure 44: Suspension Corners and Steering Clevises ................................................................. 59
Figure 45: Front Upright Jig and Partially-Constructed Left Front Upright ................................ 62
Figure 46: Tongue-and-Groove Joints for Self-Jigging ................................................................ 62
Figure 47: Rear Upright Jig .......................................................................................................... 63
Figure 48: Front Suspension Tab Jig ............................................................................................ 64
Figure 49: Link Buckling Test Results ......................................................................................... 66
Figure 50: 6061-T6 Aluminum Alloy Thread Pullout Test Data ................................................. 67
Figure 51: Rod End Shank Buckling Test .................................................................................... 68
Figure 52: Tensile Strength of Heim Joints Sourced Through Rebel Racing Products................ 69
ii
Abstract & Background

The purpose of this project was to design a suspension that would improve the
performance of the Cal Poly SAE Formula Electric car around a racing track. Performance
would be quantified through skidpad, slalom, and straight-line acceleration tests as well as
autocross lap times. The approach to meeting the objective was to increase the steady-state
lateral acceleration and quicken the transient response while maintaining predictable handling so
that the driver could extract maximum performance from the car.
The Cal Poly Formula Electric car has used a camber car configuration for the past
several years. This design is characterized by the tops of the tires leaning inwards towards the
chassis and was first proposed in 1949 by William F. Milliken, though it was not actually tested
until 1968 and his evaluations lasted until the early 1980s. Results were not widely publicized
so the camber car is still a relatively unknown concept.
Further testing of the concept was still necessary to fully understand the benefits and
drawbacks of cambered tires, and to compare them directly to conventional car racing tires. As a
result, the team was looking for someone to design a new suspension for the 2013 car with these
goals in mind. Maximilian Sluiter volunteered for this task and proposed the senior project
during the summer. Alissa Roland and Tristan French then joined the team.
Max was responsible for designing the suspension in CAD and did extensive background
research. Alissa and Tristan were responsible for sizing wheel bearings, analyzing the old
suspension, and doing most of the load and stress analysis before the design freeze. They also
spearheaded the manufacturing.
According to research, maximum lateral grip is achieved at a high-negative camber angle
(-40 degrees) but best longitudinal acceleration is had with no camber. The original design for
the suspension was a passive, dynamic camber system which varied the camber in order to
provide maximum grip in a straight line as well as in turns. The scale of this design was deemed
to be too large for the time, resources, and manufacturing skill of the Formula Electric team this
year. Therefore, it was decided to produce a highly adjustable static camber suspension that
could be tuned best performance as well as adjusted to allow for testing of the dynamic camber
concept.
The final design used the 5 link concept on each corner of the car, with 6 simple-tomanufacture tension-compression members per wheel. Analysis consisted primarily of pseudostatics to find forces in the links so that they could be sized for minimum weight and adequate
strength. Component testing and validation consisted of Instron testing for link buckling, link
thread pullout, link tensile strength, and rod end buckling. Testing on the suspension as a whole
will involve skidpad tests to find the maximum steady state lateral acceleration from various
camber configurations as well as acceleration tests to find the effect of camber on longitudinal
acceleration. The goal will be to find the best static camber setup and determine if a dynamic
camber system, active or passive, would actually provide a significant advantage if the
manufacturing resources were available to build it.
Fundamental Tire & Vehicle Dynamics

Tires generate the majority of the forces necessary for accelerating, decelerating, and
turning a ground vehicle (aerodynamic devices which could provide additional maneuvering
forces as well as downforce to increase tire grip will not be covered in this report). A vehicles
suspension system links the four tires to the sprung mass, which is essentially the chassis and
everything rigidly mounted to it. The term unsprung mass is applied to the tires, wheels, and
some fraction of the inertia of each component of the suspension mechanism. Those fractions are
based on the kinetic energy of the particular piece of the mechanism with respect to a single,
simpler variable usually the vertical displacement of the wheel. The unsprung mass on each
corner of the car is separated from the ground by the compliant tire, modeled as a spring of
between 100 and 350 kN/m stiffness1 and sometimes a weak damper. It is separated from the
sprung mass by the suspension spring(s), damper(s), and in some cases inerter(s).
Suspension Basics
The purposes of a suspension system are as follows:
Transfer forces from the tires to the sprung mass.
Put the tire in the correct orientation to the road for maximum grip.
Isolate the sprung mass from road irregularities.
Manage the load on each tire to provide the best grip.
While meeting the requirements stated above, the suspension should also have as little
inertia as possible. This not only reduces the total car mass but more importantly reduces the
effective unsprung mass. Reducing the ratio of unsprung to sprung mass helps keep the load on
each tire closer to optimal and improves the ride (isolates the sprung mass) from high frequency
road inputs because the highly-underdamped wheel-tire system then has a higher natural
frequency. This higher wheel hop frequency means that the unsprung mass is not excited as
much by inputs from bumps in the road. Since the unsprung masses motions are the forcing
functions for the sprung mass, less unsprung mass excitation means less disruption of the sprung
mass. The components with the largest effect on unsprung mass in most suspension systems are
the uprights (which hold the wheel bearings and hubs), the wheels, the tires, and any brake
components that may be mounted outboard. Lighter wheels and tires have the added benefit of
reducing the inertia that the drivetrain must accelerate and reducing gyroscopic forces (which are
usually not a significant factor in four-wheeled vehicle dynamics). For these reasons, lighter
wheels are one of the best upgrades for any car.
Tire Basics
Tires generate forces through their interaction with the road surface. The normal load
(weight) on a tire causes it to deform out of round and form a contact patch or footprint
which grips the road via several mechanisms. These include mechanical interlocking, whereby
the soft rubber is squeezed into small crevices in the road surface and acts like teeth in a rack and
pinion gear mesh, and chemical bonding. Deformation of the contact patch is resisted by a
distributed force arising where the rubber meets the road. This force pushing back on the tire is
1
Cossalter, Vittore: Motorcycle Dynamics, 2nd ed., page 56
transferred through the suspension to the sprung mass in order to accelerate, turn, and brake the
car as a whole. The objective of the suspension design in this report was to improve both steadystate lateral acceleration and transient response for better handling. Therefore, lateral force and
the mechanisms to generate it will be the focus of the following discussion, with less explanation
given for the source of tractive (acceleration and braking) forces since they also arise from a
deformation of the contact patch, just in a different direction.
There are two main methods to deform the tire for the purpose of generating lateral force:
slip angle and inclination angle. To understand what these terms mean, please refer to Figure 1.
Slip angle results when a tire is steered about an axis normal to the road. Inclination angle is a
tilting of the tire about an axis parallel to the plane of the road and orthogonal to both the slip
angle axis and the axis about which the wheel rotates (the axis defined physically by the wheel
bearings in the upright). In the figure, direction of wheel travel can be thought of as the
direction the car is moving (in reality it takes into account both rotation and translation and so is
a local velocity vector on the car) and direction of wheel heading is the direction the wheel is
pointing or being steered towards. Slip angle is given the symbol of the Greek letter alpha ()
and inclination angle gamma (). Inclination angle is shown measured from the vertical axis
passing through the contact point (or the center of the contact patch) to the dotted line which is in
the plane of the wheel and passes through both the contact point and the axis of rotation (rotation
is called spin in the diagram).
Figure 1: SAE Tire Axes
Milliken, William F. & Douglas L.: Race Car Vehicle Dynamics, page 62
Slip Angle
The manners in which slip angle and inclination angle deform the contact patch are
different. With slip angle, the tire is twisted. As shown in the diagram of Figure 1, this comes
about because the wheel is restrained to rotate about the axis defined by the wheel bearings, but
the velocity of the upright in which the wheel bearings are mounted does not align with the plane
of the wheel. Therefore, when a tread element (which can be infinitesimally small) makes
contact with the road it adheres and stays in place while the wheel rolls over it and the next tread
element enters the contact patch and sticks. The first tread element is now directly behind the
second, with the tread elements aligned to the velocity vector of the upright (direction of wheel
travel) rather than the tangential velocity on the bottom of the wheel (direction of wheel
heading). This means that the tire must distort in the section of sidewall and carcass between the
wheel rim and contact patch. The distortion of the tire combined with its stiffness causes both a
lateral force and a restoring moment (which attempts to remove the twisting deformation and
align the wheel heading to the direction of travel). The force and moment are reacted through the
tires grip on the road (which is the fixed reference frame) and the compliment of that force
causes the car to accelerate laterally in response. A diagram of this deformation can be seen in
Figure 2, which is from a viewpoint of an observer below a transparent road surface and with the
plane of the wheel along the x-axis. The curved line to the left of point A is the result of the
lateral stiffness of the tire carcass (in the range of 100 to 200kN/m)3 causing the deflection in the
contact patch to propagate forward. The tread element makes contact at A, proceeds through B
and H, and leaves the road at D.4
3
4

Figure 2: Tire Vertical Load, Lateral Force, and Contact Patch Deformation Distribution for Slip Angle
The diagram also shows that, due to the roughly parabolic distribution of normal force
within the contact patch and the linearly-increasing trend of deformation, the limit of static
friction is reached before the end of the contact patch. Sliding occurs in the lightly-loaded aft
portion of the patch and this region increases in area as slip angle increases until the whole
footprint is sliding. A noteworthy point is that the maximum lateral force occurs behind the
center of the contact length. The line labeled Fy is the resultant of the distributed lateral force and
therefore passes through the centroid of the area under the lateral force curve. The length t
denotes how far behind the center of the contact patch the lateral force resultant occurs and is
called the pneumatic trail. Pneumatic trail causes a moment which attempts to reduce slip angle,
aligning the tire with the direction of travel. As slip angle increases the sliding fraction of the
contact patch grows from the rear forward and the force resultant moves forward because sliding
friction is less than static friction. The lateral force resultant grows in magnitude but the
pneumatic trail decreases, causing the restoring moment to peak, usually just before maximum
lateral force, and fall off afterwards. The self-aligning moment is felt by the driver through the
steering system and gives useful feedback about how close to the limit of performance the car
is.6
5
6
For small values of , lateral force is directly proportional to slip angle and the
proportionality constant is called cornering stiffness. At larger slip angles the rate of change of
lateral force with respect to slip angle decreases to zero at the peak lateral force and becomes
negative past it, which means a polynomial curve fit is necessary for modeling the behavior of
tires at higher slip angles. Cornering stiffness is actually a partial derivative because the lateral
force produced by a tire also varies with the vertical load on it. It is possible, however, to
normalize the lateral force to the vertical force and produce a graph of lateral force coefficient.
Plots of this quantity for three different loads are shown in Figure 3. The beginnings of the
curves are linear, with a slope equal to the normalized cornering stiffness (to obtain cornering
stiffness, multiply by the vertical load). At higher slip angles the curves roll off and reach a peak,
past which the tire is mostly sliding.
Figure 3: Typical Load Sensitivity Behavior
It can be seen in Figure 3 that the peak lateral force coefficient decreases as vertical load
increases. This is known as load sensitivity. If this trend were extrapolated to zero load, there
would be maximum lateral force without the tire contacting the road so it is reasonable to
speculate that the peak lateral force coefficient reaches a maximum at some light load, different
for every type of tire, before again falling to zero with no load. Load sensitivity is a very
important concept and dictates much of the design of racing cars. Conventional racing cars aim
7
to have a low center of mass and a wide track (lateral spacing between the wheels). This reduces
the lateral load transfer which results from the tire forces arising in the ground plane but being
transferred to the mass center, which must be above the ground. This creates a roll moment that,
in the absence of aerodynamic devices, must be reacted by a redistribution of the vertical load on
the four tires. Less lateral weight transfer is desired for tires utilizing slip angle for lateral force
because the peak lateral force coefficient decreases on the more heavily loaded, outside tire,
meaning that combined lateral force capability is lost when weight is distributed unequally
across an axle. In other words, more grip is lost from the inside tire than is gained from the
outside.
Inclination
The deformation due to inclination is shown in Figure 4. Depicted are the shapes which
the contact patch of a motorcycle tire and car tire would assume when placed on a frictionless
surface, inclined, and under a vertical load. The round, motorcycle tire has more curvature (seen
in the black line running through each patch) than the car tire and this results in more camber
thrust (lateral force due to inclination). On a real road the curvature is suppressed by friction and
this produces a lateral force. The greater curvature of the motorcycle tire gives one explanation
for the greater camber stiffness of round-section tires compared to square-section car tires.8
Camber stiffness is the partial derivative of lateral force with respect to camber angle and at a
particular vertical load. Bias-ply tires with stiffer sidewalls generally have greater camber
stiffness than radial tires, which let the contact patch deform too easily under the road load.9
Camber is similar to inclination angle but its sign depends on which side of the vehicle the tire is
on. Negative camber means that the tops of the tires lean in towards the sprung mass, while
positive camber means they lean outward. An excellent example of negative camber is the
Milliken MX-1, which will be discussed later in this report and is shown in Figures 8 and 9.
8
9
Pacejka, Hans B.: Tire and Vehicle Dynamics, 3rd ed., page 77
Milliken, William F. and Douglas L.: Race Car Vehicle Dynamics, page 47
Figure 4: Shape of Deformed Contact Patch for Frictionless Road
10
Also apparent in Figure 4 is that the lateral deformation of the contact patch is
symmetrical about the y-axis (positioned halfway back from the front of the contact area). This
curvature more closely matches the distribution of normal load on the tire seen in Figure 2 and
results in less sliding of the tire on the road.11 The symmetry also results in much less (or even
negative) self-aligning moment (pneumatic trail) than for slip angle.12
Figure 5 is a clearer look at how a motorcycle tires contact patch changes shape as
inclination angle increases.
10
12
11
Figure 5: Changes in Contact Patch Area and Shape due to Camber and Tread Blocks
13
Camber thrust behaves very differently than slip angle. It has an approximately linear
relationship to both vertical load and inclination angle, even up to and beyond 45 degrees of
inclination, as seen in Figure 6. The linear relationship to vertical load is evidenced by the
normalized camber thrust data all falling on the same trend line even when the vertical load was
increased five-fold, from 500N (112lb.) to 2500N (561lb.); thus the tire tested does not show
camber thrust load sensitivity for the range of loads that would be seen on a lightweight racecar
such as an FSAE car. From the two graphs it can be seen that this lack of load sensitivity is not
simply due to the round section and construction of the particular motorcycle tested, since the
tire does show load sensitivity when subjected to slip angle deformation but does not do so when
inclined. The slip angle plots show curvature similar to that seen in the plots for car tires,
although there is no peak to them because the range of slip angle tested is not as large as for car
tires because motorcycles do not usually use large slip angles.
Figure 6: Load Sensitivity of Motorcycle Tires with Slip Angle but None for Inclination
13
14

14
Combining Inclination & Slip Angle

The linearity of camber thrust holds true even when there is a small slip angle present, as
is the case in Figure 7, where the right side plot shows it most clearly. The left side plot shows
the saturation (leveling off of lateral force) of the tire at large inclination angles and small slip
angles, with maximum lateral force occurring at lower slip angles when the inclination is
increased. In the left-side plot of Figure 7 it is apparent that 40 degrees of inclination produces
slightly more peak grip than 50 degrees, suggesting 40 degrees is possibly the optimum negative
camber angle. The maximum lateral force at 40 degrees of inclination also occurs at a reasonable
slip angle of about six degrees, similar to where most car tires (as opposed to round-section,
motorcycle tires) produce maximum lateral force without inclination. The amount of slip angle
required means that a car using 40 degrees of negative camber for maximum grip could still be
controlled by the conventional method of steering via altering the toe angle of the front wheels
(rotation about an axis normal to the road). If the maximum lateral force occurred at a nearly
zero slip angle then a more complex steering system on both front and rear would be needed to
eliminate slip angle and to control inclination in order to initiate the turn.
Figure 7: Lateral Force of Motorcycle 120/70-17 Front Tire at Combined Slip and Inclination Angles
15
The plots of Figure 7 were made at a single vertical load. A more comprehensive test of
combined inclination and slip angle was performed under the guidance of William F. Milliken
while he was working on his MX-1 camber car and the vehicle dynamics behind it. This
innovative test vehicle is shown in figures 8 and 9, showing its large negative camber angle and
round-section tires, making it look like no other car on the road. With older, lower-performance
tires (relative to modern tires) it achieved greater than 1g lateral acceleration consistently, and
with more modern Dunlop radials it achieved about 1.3g16, an impressive result for a pure test
machine with many compromises for the sake of adjustability. This was achieved with a negative
camber angle of just more than 23 degrees and a curb weight of 1,555 pounds with 52% of that
weight distributed on the rear track. Results from tests of the original MX-1 bias-ply tires can be
seen in figures 10 and 11.
15
16

Milliken, William F.: Equations of Motion, page 520
10
Figure 8: Milliken MX-1 Front View
Figure 9: Milliken MX-1 Side View
17
18
Figure 10 shows the lateral force generated by a motorcycle tire at a lighter load, while
Figure 11 is at a 66% higher load. These plots show the increased lateral grip due to the addition
of negative camber to slip angle at higher vertical loads. Camber values are reported as positive
but represent the magnitude of the negative camber on the outside wheel. The dotted line
represents the equilibrium state for a motorcycle without the rider hanging off to the inside of the
bike and does not pertain to this discussion of four-wheeled vehicle dynamics.
17
18
Photo from ultimatecarspage.com via Google Images

Photo from joefenstermaker on flickr
11
Figure 10: Lateral Force for Combined Camber and Slip Angles at Low Load (291 pounds) for 5.00-16 Bias-Ply, Ribbed-Tread
19
Goodyear Eagle Motorcycle Tires
The nonlinearity of the envelope of peak side force, particularly the decrease in grip
between 5 and 25 degrees of inclination, is interesting. Based on the trend of lateral force against
slip angle, one would expect a single peak, but the maximum lateral force increases again after
15 degrees of inclination and actually surpasses the first peak (near 5 degrees) at about 35 to 40
degrees of inclination. Since these tires had ribbed tread it may be due to the effects of tread
blocks on the contact patch shape and area, as seen in Figure 5.
In Figure 11 there is only one peak but there is a distinct change in slope and curvature
just after 25 degrees of camber. The peak grip still occurs at 40 degrees, similar to the highest
peak in Figure 10. At 40 degrees of inclination the addition of slip angle increases maximum
lateral force by more than 40%. The most significant point illustrated is that the addition of 40
degrees of inclination increases the maximum grip by 25% compared to a vertical tire.
19
12
Figure 11: Lateral Force from 5.00-16 Bias-Ply, Ribbed-Tread Goodyear Eagle Motorcycle Tires at Combined Inclination and
20
Slip Angles with Vertical Load of 492 pounds
Based on figures 7, 10, and 11, the optimal inclination for the outside tire in a turn is
about 40 degrees. It should be noted, however, that the MX-1 tires are out-dated and modern
sport motorcycle tires should outperform them handily, but the trends of camber thrust and
maximum grip should still be relevant. 40 degrees appears as optimal for more models of tire
than just those old Goodyears, so it is reasonable to expect modern motorcycle tires to peak at
around the same inclination, with the further reasoning that since motorcycles depend on camber
thrust for stability in a way that cars do not the tires will only have gotten better in terms of
camber stiffness and grip. Tests should still be done on modern rubber before constructing a
vehicle to determine which tire is best suited to a camber car, and whether 40 degrees is the best
for a tire in isolation. Whether static, negative camber equal to the optimal inclination for a
single tire is best for a camber car is not immediately clear because those figures do not show
what the inside tire is doing. To determine the optimal camber, the behavior of tires under
adverse inclination and light loads must be known.
If the trend of the maximum side force against camber angle in Figure 11 is extrapolated
and assumed to be oddly symmetric about the intercept with the lateral force axis, then the inside
tire looses as much lateral force coefficient (compared to zero camber) as the outside tire gains.
Figure 10 shows, however, that at lower loads the effect of camber on maximum lateral grip is
smaller. This behavior agrees with the way in which camber and slip angle distort the contact
20
13
patch. Camber relies on vertical force to deform the tire out of round while the deformation due
to slip angle is independent of load (though the amount of friction available to counteract that
deformation does vary). Therefore, the inside tire gains lateral force coefficient when weight
transfers off of it because the camber thrust is not fighting the slip angle force as much.
Continuing the trend of peak side force seen in Figure 10 would mean that the inside tire is not
significantly hampered by adverse camber and would likely not benefit enough from inclination
into the turn to make a variable-camber system worth its weight and complexity. The load below
which the leveling-off of maximum side force occurs is dependent on the particular model of tire
but if it is true for the case of positive camber it would mean that a static camber configuration
would corner just as well as a tilting machine such as a motorcycle, even with less than 100%
lateral weight transfer.
As found by William F. Milliken, lateral weight transfer increases the amount of lateral
force which a pair of negatively-cambered tires can produce. This is the opposite behavior to
traditional, vertical tires but is intuitive if one understands that the camber thrust of a tire is
proportional to load. With a pair of negatively-cambered tires, even rolling straight-ahead, there
will be a net lateral force if a disturbance causes one tire to be loaded more than the other since
the camber thrust on one side increases and the thrust on the other side which opposes it will
reduce. This net lateral force works to increase the lateral load transfer because of the mass
center of the system being above ground level. As described above, however, the inside tire does
not need to be completely unloaded to achieve maximum grip because at some point its lateral
force becomes oriented into the turn. More load increases the force but less load increases the
coefficient, so there should be an optimal, non-zero, load for the inside tire, unless the outside
tire gains significant amounts of lateral force coefficient with more load, which is the opposite
behavior of a normal car tire at small camber angles.
Interestingly, the maximum lateral force coefficient for a single tire at -40 degrees of
camber increases by more than 5% (from 1.05 to 1.11) when vertical load is increased by 66%
(from 291 pounds to 492 pounds). Those numbers are taking into account the tolerance on the
load measurement and so may be higher or lower but still show positive load sensitivity. This
means that the outside tire experiences a greater gain in potential lateral force than the increase in
vertical load, the opposite of the conventional car tire behavior shown in Figure 3. Even if the
increasing trend is weak, it is at least better than the decrease in lateral force coefficient with
vertical tires. The vertical loads in Figure 3 are more than double those in Figure 11, however, so
it is not conclusive whether normal tires would behave similarly to the cambered tires at lighter
loads and if the camber tires would experience negative load sensitivity at much higher loads.
The 900 pound load in Figure 3 would represent a small, road-legal sports car without
aerodynamic downforce undergoing a high-lateral-acceleration turn which was causing nearly
full lateral weight transfer.
It is clear from Millikens work with the MX-1 and his subsequent analysis of the data
that the gains in grip on the outside tire due to negative camber overwhelm the losses in grip on
the inside tire due to its adverse camber, and if the loads are in the right range for the tires the
inside tire likely gains grip. Fascinatingly, the beneficial effect of camber appears to be true even
for the hypothetical case of zero load transfer. This conclusion is based on simulations which
William F. Milliken and his associates performed after the extensive tests during the MX-1
14
program. The results of three simulations are shown in figures 12 and 13. The main point they
convey is that a pair of negatively-cambered tires can produce more net lateral force with more
lateral load transfer, while vertical wheels show the opposite trend (negative load sensitivity).
Also apparent is that the normalized lateral force is higher at -40 degrees of camber than at -20
degrees, even at zero lateral load transfer (h/T=0).
Figure 12: Simulation Comparing Effect of Lateral Weight Transfer on Total Lateral Force from A Pair of Tires With and
Without Negative Camber
It can be seen by comparing the plots in Figure 12 and in Figure 13 that the lateral
acceleration increases from about 0.825g to 0.925g when going from -20 degrees and -40
degrees, even without lateral load transfer. That is a difference of 12% and that difference
increases to 15.5% at when lateral weight transfer is added, even though that transfer of load is
only about 75% complete for the highest lateral acceleration in Figure 15, meaning there is still a
decent safety margin against overturning. This supports the theory that -40 degrees of camber is
best for lateral acceleration, despite Milliken using only -23 degrees on the MX-1, a value which
was the most the MX-1 could achieve.
15
Figure 13: Simulation of Lateral Force for Pair of Tires at -40 Degrees of Camber Showing More Lateral Force than -20 Degrees
Round-Section Tires vs. Square-Shouldered Tires

The beneficial effect of camber appears to be associated most strongly with motorcycle
(round-section) tires since large camber angles are necessary for significant gains and car tires
with their rectangular section are not suited to camber angles much greater than 5 degrees. Figure
14 shows car tire data which show load sensitivity for small negative camber angles (5 degrees
or less) and neutral load sensitivity for 10 degrees of negative camber. This data was taken at 7
degrees of slip angle, which produced nearly the peak lateral force at each camber angle. The
tires in that test were 225/70R15 car tires, which are radial tires with relatively high aspect ratios
(tall sidewalls which would be less stiff than short ones) and so would be more tolerant to large
camber angles than wider, lower-profile tires. Extrapolating the trend, one would expect positive
load sensitivity above -10 degrees of camber but car tires cannot use larger camber angles
because the tire ends up riding on the stiffer shoulder, shrinking the contact patch and putting
more pressure on it, which decreases grip. This is evidenced by the reduction in peak lateral
force coefficient at -10 degrees of camber versus -5 degrees of camber, though -10 degrees
produced more lateral force than 0 degrees at high loads and equal force at low load.
16
Figure 14: Load Sensitivity for Various Camber Angles at 7 Slip Angle
21
Figure 15 is a plot of maximum lateral force versus camber angle at high load for three
different round-section tires. This data was recorded by Albert G. Fonda, who made a sort of
tilting go-kart which had the rider lean into the turn. William F. Milliken was impressed by its
high lateral acceleration (1.2g) but wanted to avoid its shortcoming: what motorcyclists term
high-siding. If the kart was turning and encountered a lower-friction spot it would benignly
lay down and slide with increased inclination, but when it encountered a high-friction surface
again it tended to capsize to the outside of the turn and throw the rider out.22 The tires Fonda
tested for his 1956 paper Tire Tests and Interpretation of Data (published just after William F.
Milliken proposed to Cornell Aeronautical Laboratory that they patent his camber car idea) were
not for the kart, however, being much too large for it but suitable for an automobile. These tires
are not specified as motorcycle tires in the paper and it is mentioned in the conclusion that an
expanded test could include motorcycle tires at 30 to 50 degrees of inclination, leading to the
deduction that these are simply 1955 vintage automobile tires which were narrower and more
rounded in section than todays high performance rubber. Once shaved of their tread (in order
to remove the effects of tread pattern seen in Figure 5) these tires (shown in Figure 16) appear to
have a similar section to tires specifically made for motorcycles and the tires are also bias-plies,
which are generally stiffer than radials in certain modes and so see more benefits from camber.23
These car tires should therefore have behaved similarly to motorcycle tires, however the high
load of more than 900 pounds at which the tires were tested is not representative of motorcycle
loading.
21
Milliken, William F.: Equations of Motion, pages 496,497
23
22
17
Figure 15: Peak Lateral Force vs. Inclination Angle
24
25
Figure 16: Tires Used to Generate Data for Figure 9
The slip angle varies between data points in Figure 15 because the steering angle was
adjusted for maximum lateral force at the specified camber angle. The plots show that the
benefits of camber extend to tires not specifically for motorcycles, as extrapolating the trend
lines to 40 degrees of camber produces a gain in lateral force coefficient of more than 10 percent.
The plots also show that the trend of increasing lateral force with negative camber extends to the
case of positive camber when the load is high enough. Lateral grip appears to be degraded by
positive camber by about the same amount as negative camber improves it, though some tires
may show a more nonlinear trend and have a steeper reduction in side force with positive
camber, as seen in the leftmost plot of Figure 15.
24
Graph as presented: Milliken, William F. & Douglas L.: Race Car Vehicle Dynamics, page 49
Original data and graph: Fonda, Albert G.: Tire Tests and Interpretation of Data
25
Fonda, Albert G. Tire Tests and Interpretation of Data
18
Identically-sized 100/85-R10 PMT racing scooter rear tires are being used on all four
corners of the 2013 Formula Electric car as have been used previously. Being round-section, they
can support larger camber angles. The PMTs are about three pounds lighter than the smallest
Hoosier square-section racing tires available and have less aerodynamic drag than the wider
Hoosiers or taller, full-sized motorcycle tires.
The scooter tires are radials, which Milliken says shows less of the effects of camber than
bias-plies26 due to their greater lateral compliance, but being made for motorcycles should be
stiffer than car radial tires, since the lateral stiffness and camber stiffness of the tires significantly
affects the stability, safety, and performance of a motorcycle.
27
Figure 17: Lateral Force Characteristics for Various Round-Shouldered Tires
The scooter tires can be expected to behave similarly to the other round-section tires
which generated the data used to develop the camber tire model used here. Figure 17 shows that
ordinary scooter tires generally have a linear relationship between camber thrust and slip angle
26
27
19
similar to motorcycle tires. Due to the more intense use which racing scooters receive, they
would logically be closer to the motorcycle or racing motorcycle (SBK) tires than the treaded
scooter tires. The deviations of the enduro tires can be put down to their greater siping allowing
tread blocks to squirm and because of the change in number of tread blocks in contact with the
road as camber increases, as seen in Figure 5. The rear tire data for the enduro, scooter, and
racing bicycle are further from the motorcycles than the front tire data for those vehicles are.
This may be due to construction (ply orientation and composition) and rubber compound. The
slip angle data does not extend far enough to show a peak and drop-off in lateral force because
single-track vehicles do not typically corner at slip angles greater than those shown in the figure.
Square-shouldered, wide tires have an advantage in being able to have larger contact
patches, which due to load sensitivity effects increases grip. When the tire is deformed under
vertical load the contact patch is formed and the integral over the patch of the pressure on each
differential piece of area must equal the normal load on the tire. Stiffer sections of the tire bear
more pressure for an equal displacement. With wide, square-shouldered tires the contact patch
ends up with the same area as a narrower tire with the same tire pressure and stiffness would
have but the footprint is wide and short instead of circular or long and thin. The shorter patch
means there is less deformation out-of-round and so less heating of the rubber. This, in turn,
means that softer rubber can be used which increases grip even further. Lower pressures could
also be run to return to the same level of deformation out-of-round but with a larger contact area.
Because round-shouldered tires can make more use of the benefits of camber, they are
preferred for best lateral grip. Conventional, wide car tires are expected to be better for
longitudinal grip.
Longitudinal Force
Longitudinal forces (for propulsion and braking) arise, like lateral forces, from the
deformation of the contact patch. In this case the deformation is in line with the wheel heading
and velocity vector. Because there is a limit to the amount of distortion in any direction the
contact patch can support, there is a limit to the amount of force the tire can produce in any
direction. This means that the magnitude of the ground-plane force vector from the tire is limited
such that maximum lateral force and maximum longitudinal force cannot happen at the same
time. This is visually represented by a friction circle or g-g diagram, which is usually more
elliptical than circular. Such a diagram is shown in Figure 18, which is distorted by unequal
scales on the two axes. Of note is that this data was recorded from the ace driver Ayrton Senna
during practice for the Australian Grand Prix in 1987.
20
Figure 18: Friction Circle from Sennas Lotus at 1987 Australian GP
28
In g-g diagrams, the longitudinal acceleration is graphed versus the lateral acceleration.
The points are each simply the outputs from an accelerometer at the center of mass (or corrected
to reflect the accelerations at the center of mass) at an instant in time. They give an idea for how
much total performance a tire and vehicle can achieve. A tire or vehicle with a more rounded
outer envelope to the graph should be easier to drive and more forgiving. The maximum lateral
and longitudinal forces can have the theoretical advantage depending on the track, but generally
the tire with the most area inside the circle will perform best over a lap of a race track.
Static camber reduces the maximum acceleration and braking which the car can achieve
because lateral force is always being produced, even when there is no slip angle or lateral load
transfer. In Figure 10 and Figure 11 it can be seen that slip angle adds about 40% more lateral
force at 40 degrees of negative camber, but this means that 60% of maximum lateral force is still
being produced at zero slip angle.
28
21
This force is present on each tire of a negatively-cambered pair, one opposing the other to
achieve a net lateral force of zero. If the tire is producing so much lateral force then it is leaving
significant amounts of longitudinal grip unavailable. This could be a non-issue for low-power
cars which rely on maintaining momentum by maximizing cornering speed, but this still means
that a car with sufficient power would be hampered by the camber on straight sections of track.
All-wheel-drive could be used to regain some traction, but this is at the cost of extra weight and
still the camber would be outperformed by an all-wheel drive car with vertical tires.
Induced Drag
Generating lateral force by means of slip angle has a drawback in the form of induced
drag, which comes about because the lateral force produced is perpendicular to the wheel
heading and not the direction of wheel travel. Therefore, a component (sine of the slip angle) of
the lateral force is directed opposite the velocity vector, and only the remaining portion (cosine
of the slip angle) is actually useful for lateral acceleration. The component opposing the velocity
slows the car down or requires engine power to overcome, using more fuel and reducing lateral
grip on the drive wheels through the friction circle effect. Induced drag is a significant problem,
as seen by a sample calculation which appears in Race Car Vehicle Dynamics for the induced
drag on an Indy car. This force turns out to be 225 pounds, which requires 132 horsepower to
overcome at the 220 mph speed it is travelling.29
Camber thrust does not cause induced drag because the wheel heading aligns with the
wheel velocity and, thus, all the lateral force is used for turning and less engine power is needed
to maintain speed, which saves fuel and increases the grip on the driving wheels. All this could
be offset if the rolling resistance of a tire increases with inclination, however. Fortunately, this
does not appear to be the case.
The moment which opposes the rolling of a non-driven, non-braked wheel varies with the
cosine of the inclination angle, meaning that rolling resistance actually decreases with camber
angle. This is true for small inclination angles up to 10 degrees, where there is less than 2%
difference compared to setting the value of the cosine function equal to unity. This is well within
the scatter in tire data from test machines. Slip angle also causes similarly small changes in
rolling resistance.30 If the cosine trend continues to large camber angles, or even if the rolling
resistance stays nearly the same, then the elimination of induced drag stands as a great benefit to
using inclination to achieve high lateral force while reducing slip angle.
Effects of Camber on Vehicle & Suspension Design
In Figure 3 it can be seen that the slip angle at which peak lateral force occurs increases
at higher loads. This is the reason reverse Ackermann steering systems exist. Geometrically, the
inside front wheel in a turn is on a smaller radius than the outside and should be steered more.
This condition is called Ackermann steering, diagrammed in Figure 16. Because of the lateral
weight transfer and load sensitivity effects, however, reverse Ackermann produces more lateral
29
30
Milliken, William F. & Douglas L.: Race Car Vehicle Dynamics, pages 67-69
22
acceleration for conventional cars than normal Ackermann but comes at the expense of high
rolling resistance at low lateral accelerations due to scrubbing the inside tire across the road.31
Figure 19: Ackermann Steering Geometry
32
It should be noted that the geometry that Ackermann is calculated from only applies to
very low lateral accelerations, as slip angles in the rear mean that the turning center shifts
forward and is no longer perpendicular to the rear tires. When the rear tires have a larger slip
angle than the front, this is called oversteer. When the front tires have the larger slip angle the car
is said to be understeering. The slip angles depend on the cornering stiffness of the car and the
required lateral force at either end of the car. In steady-state cornering, the moments of the front
and rear lateral forces about the mass center must be balanced. If the mass center is not at midwheelbase, then one of the forces (rear if weight is rear-biased, front if the opposite) must be
larger to account for the unequal moment arms. Therefore, when the tires have equal cornering
stiffnesses, a car with a rear weight bias needs more rear slip angle to make more rear lateral
force and maintain equilibrium. When there is no rear-wheel steering the rear slip angle can only
come about by the heading of the car as a whole being yawed relative to the direction of travel.
This increased sideslip angle causes the car to corner more nose-in, or in other words it
oversteers. Reduced sideslip angle causes cars with forward weight biases to understeer (if no
other changes are made) and cars with equal front-to-rear weight distributions to be neutral steer.
A camber car should use conventional Ackermann because the inside wheels needs more
slip angle to counteract the adverse camber. Increased slip angle does result in more induced
drag but the inside wheel generates less force than the outside due to the lateral weight transfer
so the extra drag is insignificant. The need for more slip angle at low camber is seen in Figure
10, and the trend can reasonably be expected to continue to the case of inclination out of the turn.
Since slip angle is very hard to measure on a car, a measurable quantity called understeer
gradient is used to quantify understeer instead of a difference in slip angles. This gives the
31
32
23
amount of steering required on the front wheels in terms of the Ackermann angle plus a modifier
term equal to a constant times the lateral acceleration. If this is positive, the car understeers and
if negative, the car oversteers. The Ackermann angle depends only on turn radius and wheelbase,
so the understeer gradient can be measured by recording the steering wheel angle while
negotiating a constant-radius turn, such as during a skidpad test, and gradually building up speed
(and later accounting for compliance in the suspension). If the wheel must be turned into the turn
more as speed and lateral acceleration increases, then the car understeers. If the steering angle
must be reduced then the car oversteers, and if no change is required it is neutral-steer.
Understeer means that the car is more stable than if oversteer or neutral and also has a quicker
response time than an oversteer car under normal conditions. Oversteer cars have a critical
speed above which they become divergently unstable. Above this speed they (most likely)
cannot be controlled by a human pilot but much like modern jet fighter planes they would have
exceptional quick responses if computer stability augmentation was used.
Most race cars are close to neutral steer because of limit plow/spin effects. A front weight
bias brings understeer but also causes the front tires to reach their grip limit before the rears if the
tires are the same. Therefore, neutral-steer cars have the highest maximum steady-state lateral
acceleration as well as acceptable response times and favorable subjective handling (most racing
drivers do not like much understeer). Not all race cars have their mass centers at mid-wheelbase,
however. This is because oversteer/understeer can be changed by redistributing the lateral load
transfer, which is done by changing the proportion of total roll stiffness taken by the front and
rear tracks. Changing the roll couple distribution, as this is called, takes grip from one end and
adds it to the other, therefore a car that is balanced for neutral steer will achieve the most lateral
acceleration that it can for a given mass center location. A mass center forward of mid-wheelbase
speeds the lateral acceleration response at the center of mass33 because the initial yaw into the
turn causes more acceleration at the mass center, but the unequal weight distribution also reduces
the maximum steady-state lateral grip if the lateral force coefficient follows any trend other than
concave-up and increasing with vertical load. Normal load sensitivity (lateral force coefficient
decreasing with increasing load) means that the heavy end does not gain as much grip as the light
end looses, therefore the total grip decreases and maximum lateral acceleration suffers when the
weight distribution is not 50% rear (with the same tires on each corner)34 Ideally, one would use
the best tires on all four corners to maximize the grip, so simply using softer tires on one end is
not the best approach to avoid limit plow or spin.
Aerodynamic downforce can also be used to alter the oversteer/understeer balance by
applying more load to one track, thereby affecting cornering stiffness but most significantly
increasing the maximum grip and controlling plow/spin behavior. The narrower track (relative to
wheelbase and general vehicle size) of a camber car, combined with the narrower, round-section
tires reduces frontal area and drag coefficient compared to a conventional car. This makes a
camber car more fuel efficient and/or able to reach higher speeds and accelerate better out of
high speed turns for a given weight and engine power. The narrower, round-section tires also do
not obstruct under-floor ground-effect aerodynamic devices as much as square-shouldered, wide
33
34
Milliken, William F. & Douglas L.: Race Car Vehicle Dynamics, page 394,401
24
tires do.35 Therefore, the narrower area under a camber car due to the narrower track regains
some effective area through cleaner airflow (less turbulence and loss of total pressure) and
less tire obstruction.
With normal load sensitivity, an equal dynamic front-to-rear weight distribution is best
for braking performance36 if the car has brakes on all four wheels (as most cars do) but this
requires a rearward static weight bias to counteract the forward weight transfer. When the car
only drives the rear wheels the rearward static weight bias also helps acceleration because there
is more grip available and the rearward weight transfer during acceleration adds to this by taking
weight off the front wheels which are not contributing forward thrust. This weight transfer
affects the oversteer/understeer balance, however, so a road-racing car should not be designed to
transfer too much weight longitudinally.
The effect of scrub radius combined with caster angle causes a load transfer from one
diagonal to another when steering is applied. Scrub radius and caster angle definitions are shown
in Figure 20. Scrub radius is the distance from where a line projected along the steering axis
intersects the ground and the centroid of the contact patch. It is the lever arm for longitudinal
forces such as those from an impact with a bump in the road. A larger scrub radius decreases
steering effort at slow speeds because the tire is able to roll around the steering axis rather than
the contact patch twisting. Caster angle is the tilt of the steering axis in side view and is usually
used to increase a quantity called the mechanical trail, which is the lever arm with which the
lateral force from the tire exerts a torque through the steering system that tries to force the
steering wheel back to the straight-ahead position. Trail causes the steering weight to increase
with lateral acceleration and increases the self-centering effect as speed increases, which are
effects that most drivers expect and desire. Caster also adds inclination to both the inside and
outside wheel when steering is applied.
Figure 20: Steering Geometry Definitions
35
37
Milliken, William F.: Equations of Motion, page 524,525

37
36
25
The weight transfer caused by the caster and scrub radius creates either an understeer or
oversteer yaw couple, depending on how much camber is used and whether the scrub radius is
negative or positive. Because this is a couple and not a simple moment due to an unbalanced
increase in lateral force at one end of the car, there is no significant benefit to caster jacking
other than tuning turn entry behavior, and to a lesser degree steady-state, behavior to suit driver
preferences. The magnitude of this yaw moment is limited by the arm strength of the driver
unless a slow steering ratio or power steering is/are used. The sense of the yaw couple is affected
by how much the car relies on camber, since lateral weight transfer degrades the lateral force of
vertical tires but increases the grip for camber tires.
A major issue that must be considered when designing a camber-car steering system is
the kingpin inclination, shown in the diagram on the right of Figure 20. Kingpin inclination
causes the contact patch to move relative to the upright as steering angle is increased. It is easier
to visualize this as if the highly-cambered tire had been reduced to a thin, circular disc resting on
a frictionless surface and prevented from rolling. The contact patch would then orbit the tires
circumference as it was steered. Continuing with the visualization of the simplified, highlycambered tire, the steering axis, which is normally a yaw axis, becomes a pitch axis and causes
the wheel to pitch nose-up or nose-down, like an airplane wing, and causing either the
nose or tail to hit the ground. With the circular shape and camber angle less than 90 degrees
the contact patch moves smoothly along the circumference as steering angle is increased. This
motion is different from the forward/backward motions caused by scrub radius (in the absence of
kingpin inclination) because in that case the contact patch stays in the same position relative to
the upright and would stay in the same place on the circumference if the tire was on the
frictionless surface.
The motion of the contact patch relative to the upright causes a change in mechanical trail
in the case of dual A-arm or strut suspensions. The lateral force vector is perpendicular to line of
intersection between the wheel plane and the ground plane. If the contact patch moves relative to
the upright but the steering axis (defined by the strut or upper and lower ball joints) stays fixed
relative to the upright, then the moment arm which causes the lateral force to produce a selfcentering torque in the steering system changes. With symmetric camber and kingpin inclination,
the mechanical trail on the outside tire shrinks and can eventually become negative. The trail on
the inside tire becomes larger. Depending on Ackermann, static toe angle, and lateral load
transfer, the self-centering torque through the steering system will generally reduce and may
reverse. A similar effect can happen in airplanes when the control surfaces have too much
aerodynamic boost. When it happens in the roll mode it is called aileron snatch, so the reversal
of control forces at the rim of a steering wheel will be referred to as steering snatch in this
report. Cossalter gives an explanation and diagram of the trail change on a motorcycle when
steering is applied while the bike is leaned into the turn.38 Motorcycles do not exhibit steering
snatch because they use large amounts of static mechanical trail (i.e. without steering and roll at
the same time) and most importantly because they do not use large steering angles (more than a
few degrees) when leaned over far. This lack of steering is due to relying mostly on camber
thrust for lateral force and also the short wheelbases of most motorcycles. In certain situations,
such as slaloms, there is a danger of steering snatch at the apex of each turn, when the bike needs
38
Cossalter, Vittore.: Motorcycle Dynamics, pages 22,23
26
to reverse lean angle. If the slalom is negotiated too fast, there is a tendency for the rider to apply
too much steering angle if the lean angle is increasing too fast. This usually ends in a high-side
crash.
The shift in contact patch position relative to the upright is more pronounced when the
wheel, and not just the kingpin axis, is significantly inclined, so steering geometry on a cambercar requires more careful design and analysis. Steering snatch due to kingpin inclination can be
avoided on a camber car but requires a less-conventional suspension layout. Multilink
suspensions (also known as 5 link) have complete freedom of the steering geometry, but the
geometry changes when steered. If one positions the upper ball joint above the tire, then it can be
placed directly above the lower ball joint in order to remove the kingpin inclination. This
approach has the benefit of consistent geometry as the wheel is steered, but results in either a
more compliant upright, a heavier upright, or some combination of the two because of the
cantilever required. It also will limit the A-arm geometry because the wishbone will tend to
interfere with the upright/wheel unless it is designed with a more complex shape to clear the tire,
thereby requiring the wishbone to be heavier in order to regain stiffness. The Millikens did not
report steering snatch on the MX-1, which had a kingpin inclination roughly equal to its camber
angle. The MX-1 had a mechanical trail of about 2.6 inches in its final, best form and was tested
to 1g lateral on a 200-foot diameter circle, after which it was recorded that the steering centered
okay. This vague comment from the normally very meticulous Milliken suggests that the
steering forces may have gotten lighter than the 11 pounds per lateral g which he liked on the
zero-camber configuration and lighter even than the 8 pounds per g which was measured on
other high-camber tests and which he did not want to increase. The steering did not snatch,
however, and there are many factors that could have contributed to this outcome. The
combination of greater mechanical trail, less steering angle (due to the large turn radius), and the
lack of tire grip all helped the MX-1 (cornering at 1g did not cause sufficient weight transfer to
cause snatch). Another important factor that can help avoid steering snatch is the scrub radius. If
a positive scrub radius is used with large kingpin inclination then the front of the car is lifted
when the wheels are steered. This adds steering weight, most noticeable at low speeds. The
steering on the previous Cal Poly SLO camber cars did not snatch and they had larger scrub radii
(on the order of 2 inches) and significant caster (and so mechanical trail) to go along with the
kingpin inclination.
Static negative camber on the front tires causes a lateral force and yawing moment that
tend to counteract the yawing moment due to an impact with a bump in the road on one wheel.
When the wheel hits the bump the increase in normal load on the tire increases the camber thrust
on that corner and so produces a net lateral force at the front and a corresponding moment about
the center of mass. The lateral force causes more lateral weight transfer, however, and this
increases the lateral force such that the car tends to move away from the bump. The effect of
negative camber on the rear tires is therefore stabilizing since the lateral force is in the same
direction as in the front but the moment arm is reversed. Milliken notes that, from his experience
with the MX-1, static toe-out of about 10% of the static camber angle tends to stabilize the
wandering of a camber car.39
39
27
A camber car sees transient response benefits from faster lateral weight transfer. One way
to do this is by sending forces through the suspension members themselves, rather than the
springs, since the links are much stiffer than the spring on cars not using significant amounts of
aerodynamic downforce. The kinematics of the suspension can be set to achieve this load path by
placing the instant center of rotation in the correct place, which then puts the roll center at the
right height above ground.
The instant center of rotation is the point about which a rigid body (the upright) is
rotating at an instant in time. Perpendiculars to the velocity vectors on three or more points of the
rigid body can be used to find the instant center, but the analysis is usually broken up into two
planar representations of the suspension: front view and side view. An example of the instant
center in front view is shown in Figure 21.
Figure 21: Instant Center vs. True Center
40
The instant center will move as the wheel displaces on most suspensions. A swing-axle
suspension, like that shown in the lower half of Figure 21 has an instant center that is also a true
center and which does not move. There is only one suspension link in front view and the upright
is rigidly attached to it so that it is constrained to rotate about one physical point. The advantage
of a four-bar linkage like that in the upper half of Figure 21 is clear to see: it is much more
compact while having the exact same kinematics for that instant in time. It can be much lighter
and more rigid because there are no bending stresses and because the links are shorter, increasing
buckling resistance. The construction of the instant center for suspensions with tensioncompression members consists of drawing lines through those members and finding their
intersection point. This is because the velocity of the upright where those members attach must
be perpendicular to those (assumed rigid) links. The instant center affects the distribution of
forces in the mechanism; in other words which links have more loading.
40
28
When the instant center is drawn for both sides of the car, another point can be drawn.
This point is the roll center and represents the coupling point for the sprung and unsprung
masses. In other words, a force applied at the roll center causes no roll of the chassis. A roll
center is shown in Figure 22.
Figure 22: Roll Center Construction
41
The roll center moves vertically and laterally with suspension displacement since the
instant centers do as well. The roll center is found by drawing a line from the center of the
contact patch to the instant center of that wheel. What this is mimicking is the true center of a
swing axle suspension, where there is a force applied by the tire to a link with a pin connection at
one end. Because of this end constraint it can transmit only forces and no moments to the chassis
on which it is anchored. The true center is always above ground and since the contact patch is
always on the ground the load path does not align with the load. In order to transmit the lateral
load component through the slanted load path there must also be a vertical component, called a
jacking force. This is shown in Figure 23.
Figure 23: Vertical Jacking Force Due to Applied Lateral Force
42
The instant center is shown above ground but can be below ground, as can the roll center.
In that case the jacking force from the outside wheel is downwards. The jacking force on the
inside and outside wheel would largely cancel if the two tires produced the same lateral force,
but this is not the case due to lateral weight transfer, and especially on a camber car. Because of
this, the sprung mass sees a net vertical force with the direction of the jacking force of the
outside wheel (usually). This net vertical force must be reacted by an equal displacement of the
springs on both the inner and outer wheels. There can still be net compression on the outside and
41
42
29
extension on the inside, but there is more extension and less compression than if the roll center
was at ground level (meaning no jacking forces). Care must be taken to not let the sprung mass
rise too much due to the jacking forces, lest the car overturn. This can be combated by using
stiffer springs, which do not need to displace as much to absorb the jacking forces. These stiffer
springs further speed lateral weight transfer if they also contribute to roll stiffness and are not the
third spring acting only in heave (both wheels moving equally).
For the case of the roll center above ground the sprung mass moves up. The jacking force
also has a complimentary reaction at the tire which adds load and therefore counters lateral
weight transfer (when the roll center is above ground). This reduces the amount of load transfer
the springs must support, thereby reducing body roll. Because of this effect the roll center can be
used as the pivot point of an inverted pendulum consisting of the sprung mass of the car at the
end of a massless rod which is equal in length to the distance between the mass center height and
roll center height. The centrifugal inertia force is applied to the mass center and the tire forces
are applied to the roll center, causing a roll couple. If the roll center and mass center coincide
there is no roll, if the mass center is higher than the roll center the sprung mass rolls to the
outside of the turn, and if the mass center is lower than the roll center the sprung mass leans into
the turn.
Since jacking forces are transferred through the suspension links they react faster than
elastic weight transfer (through the springs). Geometric weight transfer (through the links)
therefore, can be used to speed transient response on a camber car which reacts favorably to
lateral weight transfer. The jacking effect of a roll center above ground can also be used to raise
the mass center for more total lateral weight transfer in steady-state cornering, but this rise must
be kept in check so that the car does not capsize.
William F. Milliken did not like high roll centers on the MX-1 because they caused so
much jacking that the camber was severely reduced, compromising the handling. This can be
countered by stiffer springs, however. The other reason he did not like the high roll center was
because a high roll center usually comes with a very short front-view-virtual swing arm length.
This means that the instant centers are nearer to their respective wheels and the shorter radius
means more camber change when the wheel is displaced vertically. If the front-view swing arm
length is equal to half the track width, then there is no change in inclination relative to the road
on either tire when there is pure body roll (inside and outside wheels displaced equally and
opposite). If the front view swing arm length is shorter, then there is inclination gain on both
inside and outside wheels- in other words the inside wheel loses negative camber and the outside
wheel gains it. This sounds good but contributes to poor rough road handling and increases
gyroscopic kickback through the steering wheel. A swing arm longer than half the track width
loses inclination but is better over bumps. Milliken did not like large camber gain when he
performed a turn reversal maneuver43, which means starting out turning in one direction and then
turning immediately in the other direction. Presumably this was because of gyroscopic effects as
the car rolled, but this was tamed with a steering damper and a droop-limiting device which
caused Milliken to be surprised at how well the car handled rough roads despite the short swing
arm. Optimal roll center height is still a compromise between faster lateral load transfer and
43
30
rough road handling (steering) effects, along with how stiff the springs can be made to control
jacking without being unduly harsh over bumps.
Transient Response
When a step steering input is applied to a tire there is a delay before the steady-state
lateral force is reached. This first-order delay is called relaxation length and represents how far
the tire must roll to achieve 63% of full lateral force (one time constant). According to Cossalter
the relaxation length to slip angle for motorcycle tires is generally between 0.12m at 20kph and
0.45m at 250kph, increasing slightly with load.44 He then states that the relaxation length for
camber thrust is much less, and has been shown in some tests to be almost negligible.45 Pacejka,
however writes that the relaxation length for camber is about equal to the one used for slip angle
and on the order of magnitude of the wheel radius,46 but this appears from context to be referring
to conventional car tires with squared-off shoulders.
If the response time to camber for round-section tires is negligible, then the incomplete
block diagram shown in Appendix A gives a starting point to model the transient response of a
static-camber car or a conventional car. Tire properties, vehicle mass properties, and suspension
settings must be determined before it could be used to provide transient response estimates. The
model assumes symmetrical camber gain on bump and rebound. For an operational model, it is
strongly suggested to break the car into four tires rather than front and rear axles, so that all the
nonlinear effects of load and camber can be taken into account.
The camber car responds faster because it has more paths in the block diagram which
increase lateral force than a conventional car does. For example, elastic weight transfer increases
lateral force at the rear of a camber car, beginning the transition to steady state. On a
conventional car the weight transfer decreases the cornering power and only the yaw
displacement adds to it. This means that a camber car can be set up stiffer in roll (assuming the
chassis is sufficiently stiff in torsion), speeding up the weight transfer and hastening the build-up
of lateral force at the rear sooner than a conventional car. The conventional car would either need
to be softer, which leads to issues on quick turn reversals if the yaw and roll get out of phase, or
suffer from a slower build-up of lateral force at the rear (since the weight transfer interferes with
the slip-angle-generated force). The camber-car advantage is present even if the relaxation length
for camber is similar to slip angle.
The benefit to resisting roll couple with elastic elements (springs) is that the lateral force
from the front tires can then be used to assist the rears via the rear roll stiffness percentage. The
benefit to reacting the roll couple geometrically is that this happens faster than elastic weight
transfer due to the stiffness of the suspension links and because of the positive feedback, but the
track must have started to generate net lateral force through some other means (such as the
weight transfer due to roll damping, which has less of a lag by virtue of being velocity-dependent
than the displacement-dependent spring forces) before the geometric weight transfer can take
over.
44

Cossalter, Vittore: Motorcycle Dynamics, 2nd ed., pages 58
46
45
31
Because camber cars build-up lateral force on the rear axle more rapidly than a
conventional car they reach yaw equilibrium sooner and should feel tighter on turn entry (less
oversteer) than a conventional car. A similar affect can be obtained on a car with vertical tires by
steering all four wheels. In this case the rears would need to turn in the same sense as the fronts.
Many premium automakers today employ small amounts of active rear-wheel steering. These
systems usually vary the phase of the rear wheel steering as speed changes. For slow speeds the
rear wheels turn opposite the front to enable a tighter turning radius for parking lot maneuvers.
As speed rises, the rear wheels turn less relative to the fronts until they begin to turn in the same
direction at high speeds. This speeds the lateral force response from the rear axle, but by
applying the input earlier rather than reducing relaxation length. This type of in-phase rearwheel-steering also creates an understeering moment to make the car feel more stable to most
drivers. Passive rear-wheel-steering, as is often done on multilink rear suspensions via body roll,
do not speed transient response as much as active systems do because it takes time for the sprung
mass to roll. The passive rear-wheel-steer delay has the same drawback as a passive dynamic
camber system.
Milliken writes that the MX-1 had a time to 90% steady state of 0.182 seconds and that
response time near maximum lateral acceleration was short and desirable.47 Corrections for rearend breakaway were necessary to reach maximum lateral acceleration but this was possible for
sustained periods so the car was still stable enough despite fast response times. The response
time quoted for the MX-1 on antiquated, treaded tires is only 75% longer than that for a 1993
Grand Prix car, if the example calculations in Race Car Vehicle Dynamics are based on
reasonably accurate tire properties.48 The MX-1 was therefore closer to a modern racing car in
terms of response than it was to contemporary road cars. Camber cars are therefore at least as
good in transient response as conventional cars, if not better.
Conclusions Regarding Vehicle and Suspension Design
For maximum steady-state lateral acceleration and fastest transient response, a car using
round-section tires at a camber angle of -40 degrees is best.
Track width should be as narrow as practical while still achieving benign roll center
behavior (movement during roll and bump) and the desired amount of suspension travel. Benefits
of a narrow track include improved rough-road handling due to smaller yaw moment arms for
impact forces49 and less steering input needed to negotiate slaloms (the track effectively becomes
wider). The mass center should be placed low enough that the car does not capsize except in
extreme situations but high enough to increase lateral load transfer. Track width adjustments can
be substituted for mass center height in order to reduce the total mass of the vehicle, which can
be increased if extra chassis members are required to raise the mass center. The wheelbase
should be used to set the longitudinal weight distribution. Using a mid-rear mounted powertrain
to keep driveline mass to a minimum (no driveshaft) the differential would ideally be at the very
back of the transmission so that there is no overhang past the rear axle line. This reduces the
47

Milliken, William F. & Douglas L.: Race Car Vehicle Dynamics, page 257 & 258
49
48
32
polar moment of inertia (in yaw) and speeds transient response and stability because the tires
effectively have more leverage to rotate the mass of the vehicle. The wheelbase length should
then be set via moving the front wheels forward or backward to achieve the desired weight
distribution. A longer wheelbase (for a given polar moment of inertia) increases directional
stability and reduces longitudinal weight transfer, aiding handling when simultaneously turning
and braking or turning and accelerating at the same time but hampering acceleration on a rearwheel drive car.
A good ratio of mass center height to track width to design with initially is 0.25, which is
where diminishing returns in maximum lateral grip appear in the Milliken simulations of Figure
13. This ratio results in a maximum (limited by inside wheel lift) lateral acceleration of 2g, found
2hcg
using the formula a y,max
, where ay,max is the maximum lateral acceleration in g, hcg is
the height of the mass center, and T is track width (the same units of length must be used to
measure both the mass center height and track width). This formula is derived from a front-view
free body diagram of the car as a rigid body and summing moments about the outside tire. The
moment arm for the inertial force is the mass center height and that for the weight is half the
track width. For a Formula SAE racecar without aerodynamic aids, 2g is more than the tires can
support and so gives a margin of safety against tipping over. In order to pass the 60 degree tilt
test during tech inspection, corresponding to a lateral acceleration of 1.7g, the h/T ratio must be
less than or equal to 0.285. The MX-1s mass center height is not reported but the h/T ratio was
probably around 0.13 based on the track width and lateral acceleration given and the weight
transfer in figures 10 and 11. A benefit to being able to use a higher mass center is that the
ground clearance can be increased in order to clear bumpier or steeper roads as might be found
on a rally stage on a mountain road.
Theoretically, more than 2g of lateral acceleration can be achieved with an h/T ratio of
0.25 if aerodynamic devices producing downforce are used. These could be either conventional
wings or a wing split along the centerline of the car and using actuators with a control system to
keep the load on the inside tire near zero while also preventing the car from overturning. This
active system should never create lift as this reduces the total normal force on the tires and so
also the lateral force capability. At low speeds, corresponding to tight turns, aerodynamic devices
are ineffective but in this case control-moment gyroscopes could in theory be used to maintain
weight transfer or even eliminate it for a conventional car, but this comes at a severe weight
penalty and does not increase vertical load the way that aerodynamic downforce does. A better
method for achieving very high lateral accelerations at slow speeds would be to use a fan and
side skirts to evacuate air from under the car in the manner of the Chaparral 2J.
Weight distribution for all road-racing cars should be equal left-to-right to ensure equal
cornering performance in each direction. This is even more important for a camber car because
an imbalance side-to-side will cause the car to drift left or right when the car is pointing straight
down the road. Correcting the drift requires a yaw that adds both tire-induced drag and
aerodynamic drag, decreasing efficiency and top speed.
The best longitudinal static weight distribution depends on the load sensitivity of the
tires. Using a very limited data set consisting of the two graphs in Figures 10 and 11 of the MX-1
33
tires at two different loads, maximum lateral force coefficient at -40 degrees of camber increases
by more than 5% (from 1.05 to 1.11) when vertical load increases by 66% from 291 pounds to
492 pounds, meaning there cambered tires show positive load sensitivity. More than two points
of data from the same tires are required to determine if the load sensitivity is concave up as well,
however it is promising that mass center can be moved rearward to regain some of the
longitudinal acceleration capability lost to camber without incurring a lateral acceleration
penalty. In the absence of more data, the goal should be to have an equal front-to-rear weight
distribution but err on the side of rear-heavy. This means that if different drivers weigh
significantly more or less then the most forward the mass center should be is mid-wheelbase. The
transient response benefits of a forward mass center can be replicated by the rear camber, a
chassis stiffer in the torsional mode, and more rear roll stiffness.
Round-section tires and a narrow track provide aerodynamic benefits and weight
reduction compared to wide, square-shouldered tires. According to William F. Milliken, a
camber car could have about 60% of the frontal area of a conventional car while the narrow tires
would open up more plan-view area for efficient under-car aerodynamics such as diffusers. The
slanted sides also would make the car much more stable in yaw, which is something that modern
sports-prototypes have trouble with (lifting off the ground and flipping once getting a little
sideways). Tire pressures should be set at 30psi hot initially but testing should be used to find the
pressures which result in the best handling. 30psi is the best value according to the research by
Fonda50 and is also close to the value which the Millikens found as optimal for the MX-151.
Cossalter shows camber and cornering stiffness dropping off with increasing pressure52,
however, so lower than 30psi may be better but should be very much dependent on the particular
model of tire.
Bump steer should be minimized on any suspension system so that the wheels point in
the direction the driver wants, but on a camber car there is a tendency to use large amounts of
camber gain in bump, which can cause a different kind of bump steer: gyroscopic kickback
through the steering system. On a smooth road the consequences of this can be kept to a
minimum and by using a swing arm length (virtual or otherwise) of approximately half-track so
that the camber angle changes very little relative to the road as the car rolls in a turn. If the wheel
maintains its inclination relative to the road there is no gyroscopic kickback because the spin axis
of the wheel does not precess in inertial space. On a rough road, a long or infinite swing arm
increases the stability of the vehicle by reducing the gyroscopic kickback and the variation in
camber thrust at each wheel. William F. Milliken wrote, after testing the MX-1, that a swing arm
length approximately corresponding to swing axles (meaning a length of about half the track
width) was a very good configuration. In every respect, seems a desirable configuration.53
Conventional Ackermann should be used due to the adverse camber of the inside wheel. Static
toe-out of about 10% of the camber angle can be used to stabilize the car if the driver complains
of wandering and also provides more Ackermann. Mechanical trail is not effective in this
50
Fonda, Albert G.: Tire Tests and Interpretation of Data,

52
53
51
34
situation54. Control forces at the rim of the steering wheel of about 10 pounds per g of lateral
acceleration provide the best feel in a performance car according to Millikens MX-1 tests.55
Kingpin inclination should be minimized if practical, and positive scrub radius and static
mechanical trail should be set appropriately to avoid steering snatch while keeping the maximum
self-centering force (at any steering angle) at the desired level. Positive scrub radius is advisable
even without kingpin inclination, in order to reduce the low (or zero) speed control forces.
The suspension should be as stiff as possible to speed transient response while still being
manageable over rough roads. The roll center should be above ground at all times, no matter the
combination of wheel displacements. A higher roll center speeds transient response but requires
stiffer springs to manage jacking and the loss of camber as a result. The roll center should be
below the mass center height so that there is roll couple remaining to be distributed elastically
between the front and rear tracks in order to tune the oversteer/understeer balance and in order to
allow the front lateral force to hasten the build-up of lateral force on the rear of the car. A roll
center higher than the sprung mass center would provide roll couple while also allowing a longer
front-view swing arm for better rough-road handling (less camber change) since the car would
lean into the turn and gain favorable inclination on all tires, however this high of a roll center
would mean extreme jacking forces and large amounts of tire scrub. A good roll center height to
begin designing with is at the level of the bottom of the chassis such that the roll center height is
about equal to the ground clearance. Front-view virtual swing arm length should be
approximately half-track.
When attempting to maximize the benefits from camber the drivetrain tends to become a
limiting factor. The issue of transmitting drive torque to the wheels at large camber angles can be
overcome with wheel hub motors or clever use of chain drives and gearboxes to circumvent
constant-velocity joint angularity restrictions. An example would be two, counter-rotating motors
(cancelling gyroscopic forces) with their axes parallel to the longitudinal axis of the car. These
motors would drive sprockets on the uprights via chains (located so as to minimize the amount of
tensioning adjustment required) and those sprockets would connect to a right angle gearbox
which then drives the wheel hub riding in the wheel bearings. The chains would accommodate
the small static toe angle necessary in the rear suspension. The right-angle gearbox would
incorporate a speed reduction so that the outboard sprocket could be reduced in diameter,
allowing more camber angle before contacting the ground. This would make the drivetrain able
to handle any camber angle while being mounted higher for more ground clearance and better
rough-road handling. It would also add anti-squat (if done properly), which could aid traction if
kept at a reasonable magnitude. A yaw moment would be generated if this system was applied to
a variable camber system but this could be used to advantage. Due to the high percentage of
lateral weight transfer, a limited-slip differential is necessary to make best use of the limited
longitudinal grip available by beginning acceleration as soon as possible after the apex of a turn.
This could be either a torque-biasing type, clutch & ramp type, or spool. A high-ratio torquebiasing differential would be the best option for a car with a reasonable mass center height to
track width ratio of about 0.25 because of the greater longevity of the gear mesh compared to the
clutch type, because more torque is sent to the outside wheel, and because differential wheel
54
55

Milliken, William F.: Equations of Motion, pages 515, 517, and 606
35
speed is allowed, unlike the spool. The spool would be best for cars with higher mass centers and
little aerodynamic downforce, which would have the inside rear wheel almost completely
unloaded in a turn. In such a case even a 3:1 torque bias to the outside wheel may be insufficient
if the inside wheel has almost no traction.
The drawback to an extremely high negative camber configuration is poor braking and
acceleration performance due to the inability to eliminate lateral force and so reach the peak
longitudinal force point on the friction circle for the tires. Race tracks are combinations of tight
turns, open sweepers, and straights. Aerodynamic downforce and camber can increase the speed
at which both tight and open turns can be taken, but on the straights the camber is detrimental.
The static camber car responds more quickly to steering inputs than a passive dynamic
camber car and can achieve the same steady-state lateral acceleration (assuming identical mass
properties), but the dynamic camber car would have advantages on certain circuits where quick
transitions are not necessary or do not make up a significant portion of the lap and where
acceleration and braking dominate. An active dynamic camber system could improve transient
response but would be difficult to control properly in all situations. To determine either the best
compromise for static camber angle or the benefits to a dynamic camber system the affect of
camber on longitudinal force must be known, but this data is scarce to non-existent. Friction
circle diagrams are about all that exists, but this is usually based on conventional cars without
significant camber. Keeping the load on the rear tires low with a long wheelbase and equal frontto-rear weight distribution would reduce the camber thrust and mean that the tires could operate
closer to their peak longitudinal grip coefficient even with camber, but the lighter load reduces
the total longitudinal force which would result. Without four-wheel drive there would still be a
loss in longitudinal acceleration potential, perhaps even greater than with a rearward mass center.
Braking would be improved with a rearward mass center because it would produce a more even
dynamic weight distribution and the typical four-wheel brake system can make use of this.
Ideally, the camber would change from highly negative for best cornering to zero for best
acceleration on turn exit and braking before the next turn.56 Alternatively, the static camber angle
can be reduced to strike a compromise for a particular track. Both of these methods were
considered when designing the 2013 Formula Electric suspension.
Dynamic Camber
Active vs. Passive
An active suspension which can change the camber angle to the optimal setting for
cornering and for acceleration or braking would have fast transient response, high steady-state
lateral acceleration, and high longitudinal acceleration. The challenge is in the control system. If
simply based on steering angle then corrective steering inputs to arrest spins would then reduce
lateral acceleration by reducing camber or even reversing it if the camber was asymmetrical
(both inside and outside wheels leaning the same way). A combination of steering angle, brake
position, lateral acceleration, and throttle position may be the best method of control. If the
56
36
steering angle is reduced and throttle dropped, no action is taken. The control system would be
looking for a reduction in lateral acceleration below a certain amount, throttle or brake
application, and steering angle reduction all at once. The difficulty is still in discerning corrective
inputs for a slide or spin versus accelerating out of a corner. GPS correlation of track position
and turn proximity may be the ultimate answer, but this is not as versatile as a passive system.
A system that leaned all tires into the turn would also have likely feel odd unless
calibrated very well since the sideslip response would occur too quickly in relation to the yaw
response expected from a normal car. A system which varied the negative camber from zero to 40 would not have this issue, since no net lateral force would be produced by the active
suspension, it would simply increase the cornering limits.
Varying the camber relies on the relaxation length for camber thrust being negligible in
order to achieve fast transient response. An active system could counteract this somewhat with a
higher camber angle to steering angle gain, but a passive system is fundamentally hampered by
roll inertia. Passive dynamic camber has the tires vertical when there is no lateral acceleration
and uses centrifugal force (initially from slip angle, then aided by the camber) to roll the car and
generate favorable camber on the tires through suspension kinematics. This requires time for the
roll angle to establish itself and so has a much longer response lag, even if the relaxation length
is short for camber. A passive system does not need a complicated control system, however, and
in certain situations it performs well. Examples of these cases would be fast, sweeping turns
where the control inputs are smooth and relatively low-frequency. Here the suspension does not
have the problem of the roll and yaw getting out of phase, as would be the case in a slalom.
There is also a benefit to the passive system when correcting a spin, since the camber does not
change as fast as the steering input. The main drawback, then, is in turn reversal. Too high a
frequency causes the car to be leaning out of the turn and so poorly cambered. If the frequency is
high enough and the motion well damped then this problem can be kept to a minimum since the
amplitude of the camber will be small and the weight transfer small as well (assuming zero
geometric roll resistance so that there is maximum roll couple to be used for steady-state roll
angle and camber thrust). Depending on how much of the track is made up of fast, wide-open
sweepers, the passive setup may be fine, as long as there are not many quick turn reversals
required.
Both the active and passive dynamic camber systems will be heavier than simple static
camber setups and this must be taken into account when the gains in longitudinal grip from
camber reduction are found through testing. Weight increases for the passive system due to the
increased complexity of the suspension mechanism and drivetrain (large amounts of axle plunge)
compared to a static camber setup and mass is added to the active setup by the actuators, control
system, power source, and drivetrain.
A system under manual control and separate from the steering would theoretically
accomplish the desired effect but would put too much work load on the driver if the camber
switch was anything other than a bi-polar one: vertical or full negative camber and only to be
activated before or after a corner, not entering or exiting a turn. In those transitions the friction
circle would not be fully exploited as it would be in an independently-controlled, active system.
37
Polynx
The initial suspension concept was a passive dynamic camber system which has the tires
vertical when not turning but which generates a large negative camber angle on the outside tire
and a small positive camber angle on the inside tire when turning. The difference in camber
angles is in recognition of the fact that the lightly-loaded inside tire does not benefit from camber
as much as the heavily-loaded outside one. A single spring and damper unit is installed on each
corner and an additional spring and damper connects the inside and outside wheels on each axle
in order to provide additional spring rate in ride and pitch. This minimizes the pitch angle and
therefore keeps the wheels more vertical during acceleration and braking so as to maximize
longitudinal force.
The desired camber change happens in response to the opposing displacements of the
wheels when the sprung mass rolled, similar to the effect of a front-view swing arm length of
less than half the track width. The issue with using a conventional suspension (such as dual Aarms) with a short swing arm length is that it usually comes with a high roll center, which
reduces the roll couple and so the roll angle which the suspension relies on (though a higher roll
center would increase the natural frequency in roll and so reduce the response time). The other
issue with a conventional suspension design was that it was impossible to achieve the highlyunequal camber change between bump and rebound travel. As a result, a novel new suspension
mechanism was devised.
This is a system incorporating a four-bar linkage with an extended coupler link forming a
rocker arm controlling the path of the lower ball joint of the upright. Connecting to the upper ball
joint is a link that connects to a floating node, which is a pivot on the end of a link that has its
ground pivot point on the chassis. A third link connects to this floating node, but this third link
has its other pivot point floating as well. This floating pivot point is shared with the lower fourbar linkage and is also the floating fulcrum for the rocker arm controlling the lower ball joint.
This suspension falls into the category of VXI suspensions because it uses two floating pivot
points and a camber link connecting to the upper ball joint. Figure 24 shows a basic sketch of
the general concept but with equal inclination on the inside and outside tires rather than the
approximately vertical inside tire that was deemed superior because it reduced the amount of
plunge necessary in the driveshafts.
38
Mechanism Design
Figure 24: General Polynx Design
In order to minimize bump steer, an intermediate crank is used so that the steering
linkages follow the suspension links centers of rotation as camber changes. This crank is
anchored to one of the suspension links which connect to the chassis. There is an inner tie-rod
connecting one of the crank arms to either a steering rack or a lever at the end of the steering
column (like a go-kart). Another tie-rod connects the other crank arm to the steering arm on the
upright. This system eliminates bump steer in a straight line but not when the wheels are steered.
This suspension design was given the name Polynx as a reference to both Cal Poly and
the number of links in the system. Polynx is a planar (in front view) mechanism with three pivot
points on the chassis, but for certain kinematic profiles two of those pivots can be in the same
spot, resulting in two rocker arms in the suspension, reducing the number of joints and increasing
the rigidity. The three-pivot version allows more complicated wheel paths such as short-travel,
high camber change ones, but can also be used to achieve a high-travel suspension with little
camber change. The two-pivot design provides a convenient position for a direct-acting coil-over
shock absorber (coil spring concentric over a linear damper). Figure 25 shows a CAD rendering
of the steering system on a 2-pivot Polynx system but without the spring-damper unit installed.
39
Figure 25: Steering Mechanism Shown on a Simplified (2 Pivot) Polynx Suspension
The roll center of the Polynx configuration chosen for the FSAE camber car is fixed
relative to the chassis and is at ground level at static ride height in order to increase the steadystate roll angle without causing jacking of the sprung mass or lateral tire scrub.
With the extreme camber change and low roll center the driveshafts must change length
significantly. One solution would be wheel-hub mounted electric or hydraulic motors. What was
settled upon, primarily due to familiarity, was a system using conventional, Rzeppa-type
constant velocity joints and a shaft that can change length by means of ball-bearing splines.
Driveshafts were also to be used in the front for inboard brakes because the current
Formula Electric wheels are too small to fit the brakes inside them and larger wheels would be
heavier and require heavier, larger tires. Mounting the brakes on the inside of the uprights, as
was already done on the previous Formula Electric cars, would interfere with linkages. These
driveshafts will add weight and so cancel the weight advantage of the small motorcycle tires, but
the inboard brakes relieve the suspension links of the need to react brake torque and so they can
be made lighter. Full-size motorcycle tires would be large enough in diameter to fit brakes in and
would allow the camber angle to be shared better between the inner and outer constant-velocity
joints, therefore permitting more camber. Running inboard brakes does show that a four-wheeldrive Polynx version is possible. The main drawback of the inboard brakes is a smaller
maximum steering angle, but this should only affect very tight parking maneuvers which are not
important for track performance. This version of Polynx would be better suited to vehicles larger
than FSAE cars, where the weight penalty of tires large enough to fit outboard brakes in the front
would not be such an issue. Inboard brakes are a logical choice in the rear because the drivetrain
is already there for propulsion.
Design Direction Change
As the time to manufacture the Polynx suspension drew near, it was decided that the
Formula Electric team did not have the resources to manufacture the Polynx suspension for the
2013 competition. Additionally, the team did not wish to redesign the electronic stability control
system to account for the inherent torque-vectoring that the dynamic camber system included
40
due to the change in rolling radius when the tires lean. Along with the complex drivetrain
(plunging shafts and high-angle constant-velocity joints) necessary for Polynx, the necessary
changes to the electronic yaw & traction control system and the manufacturing difficulty
prompted the decision by the team to abandon the Polynx suspension system and adopt an
approach of optimizing the much simpler static camber setup.
The suspension team still desired to know more about camber and the specific
characteristics of the small motorcycle racing tires being used (due to their light weight).
Therefore, the design goals became to design a suspension system that could accommodate any
camber angle between approximately -30 and +6. This would enable testing of many different
camber configurations so that the optimal static camber angles on the front and rear could be
determined for the competition. The second main goal of the adjustability was to simulate the
Polynx suspensions steady-state cornering grip by performing skidpad tests with asymmetric
camber of -24 on the outer wheels and +2 on the inside wheels.
Final Design: Static Camber

Kinematics
The front-view swing-arm length was set at approximately half the track width, in
accordance with Millikens recommendations. The roll center static height in the front was set at
about 2 inches above the ground. This was largely determined by the minimum clearance
between the inner suspension pivot point and the ground, as well as the necessity for the
suspension links to clear the brake rotor. The rear roll center height is about 0.25 inches higher
than the front to speed transient response by having more geometric weight transfer in the rear
than the front. The front-view suspension pivot points are shown in Figure 26 for the front
suspension and in Figure 27 for the rear suspension. Dimensions are in inches and the solid,
horizontal, black line is the ground level. The dotted, vertical line is the centerline of the chassis.
The shorter, dotted lines represent the springs and connect to triangles which show the rocker
pivot geometry. The four lower points are the inner pivot points for the two pseudo-A-arms.
The steering link inner pivot (on the steering rack) is positioned 0.012 inches lower and
0.114 inches further outboard than the inner pseudo-A-arm pivot points. This was the best
position for minimum bump steer as determined by SolidWorks after the upright geometry had
been set for simplicity of manufacturing, and with the knowledge that the steering link had to
clear the brake rotor and the steering rack had to be mounted low in order to allow the drivers
legs to pass over it.
41
Figure 26: Front Suspension Pivot Point Front View Geometry
Figure 27: Rear Suspension Pivot Point Front View Geometry
Both the front and rear suspensions have only one spring per corner (no heave springs)
and there are no anti-roll bars because they would add weight and complexity and their purpose
can theoretically be fulfilled by changing spring rates.
The static camber angle is -22 degrees on all four wheels because it is near the limit of
the constant velocity joints being used. Joints that would allow higher angularity would be
heavier, and more complicated drivetrain layouts were rejected by the drivetrain team as too
much work.
42
There is no anti-dive or anti-squat built into the suspension because the undamped
oscillations caused by the coupling of longitudinal and jacking forces through the suspension
members tends to reduce the braking or acceleration capabilities.
The side view geometry of the front suspension is very important because it sets the
steering geometry. The pseudo-A-arms define upper and lower virtual ball joints at the
intersection points of lines projected along the suspension links in those pseudo-A-arms. In the
rear the side view geometry is set in order to reduce the maximum forces in the links between
braking/driving and turning. This is accomplished by adjusting the splay angle of the links. Side
view geometry is shown in Figure 28 for the front suspension and Figure 29 for the rear.
Figure 28: Side View of Front Suspension Points
The vertical centerline is the center plane of the rocker arm and coil-over spring mount.
The solid, horizontal line is represents the ground level. The front pushrod is in the plane of the
rocker arm when the wheels are pointed straight-ahead but the rear pushrod is slanted because
the rocker arm had to be moved rearward to clear drivetrain components.
43
Figure 29: Side View of Rear Suspension Points
Upright geometry is more complex, especially on the front where the clevises are angled.
The front view geometry of the front upright is shown in Figure 30 and the front view of the rear
upright is in Figure 31.
Figure 30: Front View of Front Upright Geometry
44
In the front view sketches of the uprights and tires the wheel offset of +0.45 inches can be
seen, as well as the 0.75 inch thickness of the hub and the 0.158 inch or 0.177 inch protrusion of
the inner bearing race from the outer face of the upright. In Figure 30 the angle of each clevis
block is shown (24 degrees for the bottom and 6 degrees for the top), as well as where the
pushrod clevis spherical bearing is located (the 1.296 inch line at the top of the upright). The
0.75 dimension is the distance between the two plates of the clevis (in other words, the throat).
The solid, horizontal line is the ground level in both drawings.
Figure 31: Front View of Rear Upright Geometry
The side view of the uprights is harder to show because of the three-dimensionality of the
parts. The lower pivot point shown in Figure 30 is only for the pseudo-A-arms. The steering arm
is mounted in the same clevis but is not in line with the lower pivot points. This geometry is seen
in Figure 32, and the construction of the upright will be discussed in a following section.
Figure 32 is a top view of the left front uprights lower clevis block, but the part is the
same on the right side, just flipped over so as to be mirrored. The part is also the same for the
high and low camber parts. The offset of the bolt hole for the steering rod end (0.906 inches) is
necessary for the correct Ackermann with a front-mounted rack. The front-mounted rack is
necessary to allow a one-piece steering column, and a rear-mounted rack would have required
multiple universal joints or a right-angle gearbox. This is a common issue, characteristic of the
small size of FSAE cars.
45
Figure 32: Top View of Left Front Upright Lower Clevis Plate
Figure 33 shows the top clevis block for the high camber front uprights. The upper and
lower plates of the top clevis are different from each other and are different for the high and low
camber configurations because the length of the plate must be adjusted to meet and weld to the
outer face of the upright when the angle of the top clevis changes for high or low camber.
Figure 33: Top View of Left Front, High Camber Upright
46
The 1.5 inch spacing between the bolt holes for the pseudo-A-arm rod ends in the top
clevis, as well as the relative location of the hole for the rotating clevis bushing tube (the large
hole in the middle) do not change between high and low camber parts.
The four spherical bearing center points are symmetrical about a plane perpendicular to
the clevis plates and containing the wheel spin axis. This plane is also perpendicular to the road
when the wheels are pointed straight-ahead, meaning the upright is not leaned forward or
backwards in side view. Caster is achieved entirely by the top-view angles of the suspension
links.
The side view geometry of the rear upright is easier to show, but the rear hub and
constant velocity joint are not normal to the plane containing the suspension points. As a result,
the vertical spacing shown in Figure 34 is not exactly to scale. The fore and aft location of the
spherical bearing center points relative to the vertical centerline is correct, though. This
centerline passes through the axis of wheel rotation. The view is once again from the left looking
right, with the front of the car being out of the picture frame to the left of the page. The toe link
connects to the forward-most point and the pushrod to the rear-most. The other points are for the
pseudo-A-arms.
Figure 34: Side View of Rear Upright Geometry
The theoretical center-to-center lengths for each link are shown in Table 1, calculated
using CAD and the desired steering geometry and kinematics.
47
Table 1: Link Lengths
Link Name
Front Lower Rear
Front Lower Front
Front Upper Rear
Front Upper Front
Front Pushrod
Steering
Rear Lower Rear
Rear Lower Front
Rear Upper Rear
Rear Upper Front
Rear Pushrod
Rear Toe
Center-to-Center Length (inches)

-22Camber
0Camber
13.030
13.368
12.591
12.941
8.277
12.433
8.861
12.829
8.134
10.136
13.932
14.286
12.629
12.833
12.629
12.833
8.469
12.574
8.469
12.574
12.386
11.735
12.369
12.577
The multilink system allows much more freedom in steering geometry than dual A-arms
do. This is because the steering axis is defined by the instantaneous centers formed from the two
pairs of lateral control arms, rather than constrained by two physical pivot points on the A-arms.
This allowed the scrub radius to be reduced significantly from the previous suspension despite
the wheels and tires being too small and narrow to house any steering or brake components. The
previous (A-arm) suspension had more than 2 inches of scrub radius and about 30 degrees of
steering included angle which combined for significant steering kick-back and also lifted the
front of the car an amount equal to about 50% of the ride height, shifting weight rearward when
steered. The multilink has a scrub radius of approximately zero with a -22 degree camber
configuration with a steering included angle approximately equal to the camber angle. Front-end
lift has been reduced to nearly zero, partially due to the placement of the pushrod on the top of
the upright. As the upright turns the pushrod changes its angle and effectively shortens which
cancels-out the effect of the rising front upright so that the sprung mass remains at the same
height. Self-centering on the new suspension is achieved via mechanical and pneumatic trail
rather than the front-end lift used previously and this means steering weight changes are in
theory more dependent on speed and lateral acceleration.
According to research, the inside tire must be turned about 6 degrees more than the
outside one in order to produce maximum grip with a large negative camber angle. To achieve
this, a goal of 100% geometric Ackermann was set. About 80% was achieved because more
resulted in too much nonlinearity on the inside wheel, causing structural issues (large loads in the
links) and link interference. A graph of the Ackermann is shown in Figure 35. The steering angle
of the road wheels were calculated using CAD. The toe angle was defined to be the angle
between the straight-ahead and the line formed by a plane containing the circumference of the
tire intersecting the plane of the road.
The amount of Ackermann shown in the graph is supplemented by the static toe-out of
about 2 degrees on each wheel for stability. Therefore, the effective Ackermann is closer to
100% than it appears. In the rear, the static toe-out can be set to produce nearly the optimum slip
angles on both the inside and outside tires for the most common turn radii while still being close
48
to 10% of static camber as recommended by Milliken. When slip angles develop in the rear, the
turning center shifts forward and the theoretical Ackermann no longer applies.
60
Outer Wheel Angle
Road Wheel Angle (deg)
50
Inner Wheel Angle

Ackermann Inner Angle
40
30
20
10
0
0
15
30
45
60
75
Steering Wheel Angle (deg)

Figure 35: Ackermann of the Front Multilink Steering System
The steering links outer pivot was positioned on the front of the upright so that the rack
could be mounted in its forward location without forcing the steering link to cross over the lower
lateral control arms. Crossover is not possible because the outer steering rod end must be
mounted low in order to clear the brake rotor. The inner steering arm location which was found
to have the least bump steer with the desired upright configuration (steering arm mounted low as
part of the pseudo-A-arm clevis) was determined using front-view sketches of the front
suspension in its maximum bump, maximum droop, and static positions. This location
necessitated a rack and produced the bump steer curves shown in Figure 36, which shows that
there is practically zero bump steer in a straight line. The short outer steering arm length means
that the inner steering link pivot point does not need to move very far to achieve large steering
angles. That then reduces the bump steer when the wheels are turned.
The steering ratio of about 2 (two degrees of steering wheel turn per degree of road wheel
steering) means that the drivers arms do not cross-over each other, even in the tightest turns, and
that he or she does not need to reposition his or her hands on the wheel. This is safer, more
comfortable, and can help the driver go faster around a corner because of the enhanced ability to
correct spins compared to a slower ratio. In order to keep the steering weight manageable, the
mechanical trail was set at a value of about 0.4 inches when the wheels are pointed straightahead. This resulted in an expected steering weight of about 11 pounds per g of lateral
acceleration for large-radius turns, measured at the rim of the 9.375 diameter steering wheel. A
benefit of the fast steering ratio is that the steering rods inner pivot point moves less for a given
turn radius than it would with a slower ratio. This reduces the bump steer when negotiating turns
49
on a rough road, improving stability and the speed at which the turn can be taken. The steering
weight is similar to that on the MX-1 which William F. Milliken found satisfactory.57
Straight-Ahead
Outside Wheel, 8.2ft Outer Turn Radius
Inside Wheel, 4.2ft Inside Turn Radius
1.0
0.8
Bump Travel (inches)
0.6
0.4
0.2
-0.5
-0.3
0.0
-0.1
-0.2
0.1
0.3
0.5
-0.4
-0.6
-0.8
-1.0
Toe-Out (degrees)
Figure 36: Bump Steer for Multilink with Rack & Pinion Steering
Due to the complex geometry involved, a full analysis of steering force variation with
steering angle and lateral acceleration was not completed in the short time available for the
suspension re-design after the abandonment of dynamic camber and before manufacturing had to
begin. Since the kingpin inclination was similar to the MX-1 and the previous Cal Poly cambercars it was believed that the steering would not snatch, although the steering weight would likely
decrease at high lateral acceleration. To add more self-centering in those high-g situations the
brake calipers were placed on the front of the upright so that their inertial forces exert an
understeering yaw moment on the road wheels. This did not prove to be enough and the steering
snatch at large steering angles during testing, however it is not known how much steering angle
was applied when the snatch (very abruptly) occurred because the frame geometry is not exactly
as-designed so the drop-down clevises do not act as steering stops as they were supposed to.
Therefore, it may be that the steering is going too far, a conclusion seemingly supported by two
failures of the front suspension (which will be discussed in the design verification section),
although improper handling of the rod ends during construction is also suspect. The conclusion
from the initial test drives is that the large steering angles, negligible scrub radius, and small
57
50
static mechanical trail combined to cause the steering snatch. The scrub radius which was
thought to be a good feature because it reduced the front-end rise with steering also increased the
zero-speed steering weight to unacceptable values. This shows the importance of considering the
interactions between all the steering geometry factors in order to design a good steering system.
The large steering angle is still desirable, so other measures should be taken on a future
version of the suspension in order to eliminate steering snatch. A larger, positive scrub radius
(perhaps 1 inch) and smaller steering included angle would be the most important changes. This
could be achieved by moving the lower pseudo ball joint inwards, accomplished by increasing
the plan-view included angle of the lower pseudo A-arm links, specifically by moving the lower,
aft inboard pivot point rearwards on the chassis and leaving the lower, forward pivot point where
it is. An additional benefit to this change in geometry is that it would reduce the peak load in that
suspension member, which occur during braking and when cornering at high lateral accelerations
with full steering lock. To reduce the kingpin inclination even more, the spacing of the pseudoA-arm rod ends in the upper clevis of the front upright should be increased, along with the
spacing of the inner pivot points of the upper pseudo-A-arm, thereby positioning the virtual
upper ball joint further outboard. Offsetting the upper and lower clevises in the upright, and the
upper and lower inboard pivot points may also allow the mechanical trail to be made more
consistent throughout the range of steering travel. The geometry of the steering arm on the
upright and possibly the rack position will need to be modified to keep the correct Ackermann
and steering ratio.
With the multilink design the brake rotor does not necessarily need to be positioned on
the inner face of the upright, but since the caliper would not fit in the wheel, the spindle would
be severely cantilevered out from the bearings and so would need to be stronger and heavier to
handle the bending stresses, and the bearings would need to be larger. Therefore, weight is
reduced by locating the brakes on the inner face of the upright (as seen in Figure 37), which is
where they were on the previous suspension. This outboard brake configuration eliminates the
complex plunging shafts necessary for the front inboard brakes on the Polynx design, but means
that the bottom front links are very highly stressed while reacting the brake torque in addition to
the longitudinal force.
The motion ratios for the front and rear spring systems are slightly less than 1:1, meaning
that the spring and damper compress slightly less than 1 inch per inch of upward wheel travel.
This ratio was chosen to provide the most leverage for the spring without exceeding the limits of
the damper stroke. The leverage is important to provide a stiffer suspension so that it does not
bottom out. This could also be achieved with stiffer springs but it was desired to save money by
using springs already in inventory. The spring rates are equal front and rear at 200 pounds per
inch because the weight is distributed approximately equally between the front and rear axle. A
slightly higher rear natural frequency helps provide a level ride and increases driver comfort. An
anti-roll bar could be used on one end to tune the understeer/oversteer balance through changing
the roll couple distribution without affecting the ride. If the weight distribution was slightly rearbiased this would not be necessary because the stiffer springs which provide a higher rear natural
frequency also increase the rear roll stiffness and balance the handling. Geometric anti-roll must
also be taken into account when choosing spring rates.
51
Ride height is 1.25 to the bolt heads for the front lower suspension points which are the
lowest point on the car. Rake angle is zero. The low ride height is necessitated to reduce the
angle of the outer constant-velocity joints. A higher ride height was desired in order to increase
the mass center height (and so the lateral load transfer which camber relies on) along with
improving performance over rough roads by preventing the chassis from scraping the ground
over bumps.
The desired weight distribution for the car was 55% rear to provide a good compromise
between lateral grip and longitudinal grip but the actual weight distribution turned out to be
slightly less than 50% rear when the driver was in the car. The wheelbase of the car shrank from
66 inches to 62 inches because the chassis was longer than necessary in the midsection and it
was not desired to move the front wheels forward along with the rear because it was believed
that would result in too rearward of a weight distribution while increasing the turning radius. In
hindsight, extending the wheelbase via moving the front wheels forward would have been better
because it would have improved driver compartment ergonomics as well as returning the weight
distribution to the desired value. The track width was set at 48 inches based on a projected mass
center height of 12 inches which turned out to be much higher than the actual value, which was
about 8 inches.
The tires are PMT rear tires for small racing motorcycles. The size is 100/85-10 on all
four corners. These are what have been used previously on the Formula Electric car and what the
current wheels fit. They are lightweight and are probably the best option in their size. Larger tires
and wheels from a full-size sports motorcycle would allow more camber but would not fit in the
2012 to 2013 budget. They would also have considerably more inertia.
Component Design & Analysis
The final design for the suspension is based on the traditional multilink (or 5-link)
concept. Six individual links constrain the upright in all six degrees of linear and rotational
freedom. In such a system four of the links replace the pair of A-arms in a conventional system.
A fifth link controls the toe angle of the wheel (and would be called a steering tie-rod on the
front suspension). The sixth link is either the spring itself or the pushrod (or pullrod) which
connects to the spring through a rocker arm. In the design chosen there are pushrods for both the
front and rear suspensions because the spring-damper units are too bulky to attach directly from
the chassis to the upright without interfering with other components. Each of the six links per
corner are simple, aluminum-alloy rods with right-hand female threads on one end and left-hand
female threads on the other so that two heim joint rod ends can be threaded into each end and the
center-to-center length of each link can be adjusted by loosening jam nuts and rotating the rod in
one direction or the other without removing the heim joints from their clevises. For large changes
in camber (which would exceed the limitations of the threaded lengths of the rod-ends), new
aluminum rods can be made much more quickly and easily than welded steel A-arms, which was
what the previous suspension used. Figure 37 is a CAD rendering of the full suspension in
isolation. The left side wheels have been removed to show the uprights and hubs better.
52
Figure 37: Final Suspension Design
In the rear, the heim joints are oriented with their axes of full rotation oriented
longitudinally so that the camber can be adjusted throughout an extremely large range without
requiring changes to the upright or chassis. The rear links are splayed, forming a V in plan
view, similar to an A-arm, in order to better react braking and drive forces. The consequence of
this is that the heim joints are very near their maximum misalignment angle but this is acceptable
because only a few degrees of static toe-in or toe-out are necessary for tuning the stability and
response and both ride motion and camber adjustment are accommodated by the heim joints
axis of infinite rotation (about the bolt axis). The right rear upright and suspension links are
depicted as-manufactured and unpainted in Figure 38.
Figure 38: Right Rear Suspension Corner
The pushrod from the rear spring and rocker is attached to the bottom of the rear upright
because it allows the pushrod to be more in line with the resultant force vector from the lateral
tire force and the weight than if it attached to the top of the upright. It also allows a lower spring
and damper placement which means that the rocker arm mounts can share a chassis tube with the
suspension points and therefore reduce the complexity and weight of the chassis. Locating the
pushrod on the bottom of the upright also means it does not need to change length significantly
with large camber changes, as would be the case with the pushrod mounted on the top of the
upright. What this means in practice is that fewer rear pushrods need to be made since the
turnbuckle adjustment can cover a large range of camber. By having the pushrod acting on the
upright through its own clevis, rather than acting on an A-arm as on the previous suspension, all
53
the suspension links are two-force members in theory, and can therefore be lighter and/or have a
higher factor of safety.
Both front and rear coil-spring-over-damper pairs are horizontally-opposed, as seen in
Figure 37, so that forces due to brake dive, acceleration squat, and heave (all four wheels moving
up or down equally) are canceled within the spring-damper unit mount and putting less stress on
the chassis. These mounts each consist of two metal tabs which form clevises for the spherical
bearing in each spring-damper unit. There are two holes in each tab, one for the left side springdamper units bolt and one for the right side unit. The net forces resulting on the chassis at that
point are lower than with separate mounts or with spring-damper units that are not directly
opposed to each other and those forces only come about as a result of the lateral load transfer.
The rockers are all oriented with their rotational axes longitudinal, and the springs act on the
same side of the rocker as the pushrod does, therefore the pushrod and spring forces partially
cancel (more or less, depending on the lever ratios for the pushrod and spring) and reduce the
stress on the rocker arm pivot.
In the front, the heim joints must be oriented with their rotational axes vertical in order to
accommodate steering. This then creates challenges when changing camber because this must be
accommodated by the limited misalignment capability of the rod-ends. With the chosen front
view swing arm length of approximately half track and roll center height at the center of the
bottom chassis tube the pivot points fixed to the chassis do not need to change position or angle
in order to retain similar kinematics when varying camber. The clevis angles on the upright do
need to change angle, however, and this led at first to a three-piece design for the front upright,
with a central section serving as a brake caliper mount and a locator for the wheel bearings, an
upper clevis piece where the top two suspension links would mount along with the pushrod, and
a lower clevis piece which would mount the bottom two suspension links and the steering link.
There would be three bolt holes in the center section for each clevis piece. Two through-bolts
would pass through both the clevis piece and center section, occupying two of the holes. One
bolt would serve as a pivot point and in order to change camber the second bolt would be
removed and inserted into the third hole in the upright and clevis piece. This would change the
angle of the clevis and the location of the outer pivot point. The change in location of the pivot
point was beneficial to steering geometry (with the initial steering rack placement) and frontview kinematics. Only two angle positions were necessary to cover the desired range of camber
when using relatively compact heim joints.
The three-piece design was abandoned when the steering system was finalized. A major
goal was to use a single-piece steering column (rather than one or more universal joints) to
reduce the backlash and deflection in the steering mechanism as much as possible. To achieve
this, the steering rack had to be mounted very far forward do to the short wheelbase (required for
acceptable weight distribution). The initial rack placement behind the front wheel centerline
would have required a gearbox or several universal joints. With the front-mounted rack, the
steering links outer pivot point had to be much closer to the tire than the other suspension links
pivots in order to achieve the desired Ackermann effect. The result of this was that when the
lower clevis was changed to the low-camber configuration the steering arm would hit the tire.
Since the same clevis piece could not be used for both configurations, it was decided to move to
a single-piece, sheet-steel, welded upright that would be lighter and stiffer than the three-piece
54
one. Two different uprights would be required per side, however: a high-camber one and a lowcamber one.
The biggest issue that came up before manufacturing was that the rack was several inches
too wide for the desired inner steering link pivot point. There was not enough time or budget to
obtain or make a new rack, and changing the steering geometry so significantly was
unacceptable. The suspension points could not be changed because the chassis subsystem team
did not wish to change the floor of the chassis to be wider or to bend around the rack (the frame
rails had to extend further forward for driver leg protection, a problem common to FSAE cars but
not on full-size racing cars). This design constraint meant that special steering rack clevises were
required to achieve the correct steering geometry. These allow the rack to mount on top of the
lower frame rail while the clevis itself extends down to the correct position. This then creates the
possible problem of the clevis hitting the frame when steering, but due to the steering ratio
selected there was sufficient clearance simply by extending the bottom control arm tabs by about
0.5 inches (meaning the frame rails are narrower while the lower control arm pivots stay in the
same spot). The clevises therefore also act as steering stops to prevent the heim joints from
binding and failing. These clevis pieces were milled manually from blocks of 7075-T651
aluminum alloy and use a circlamp design similar to a two-piece shaft collar. Four -28 bolts
clamp the clevis to the 7/8 inch diameter steel rack, and provide a sufficient level of friction to
transfer the expected loads with a healthy margin of safety when torque properly. These clevises
also circumvent the problem of the steering rack being too wide, which prevented the rack from
being mounted at the correct height with clevises on the ends and the frame going around the
rack housing. A final benefit is that the rack can be slightly further forward (and higher) that it
otherwise would be, which decreases the steering column angle and increases driver comfort.
Figure 39 shows the actual steering clevises and they can be seen in the CAD rendering of Figure
37 installed on the rack.
Figure 39: Anti-Bumpsteer Steering Rack Clevises
Ideally, the driver and roll hoop would be moved back, with the motor controller area
shortened. This would allow the chassis proper to end just ahead of the front suspension
55
attachment points, permitting the steering rack to be positioned at any height without interfering
with the drivers range of motion. Before setting the width of the suspension pivot points, the
rack would be chosen (if a rack was to be used) so that the problem of the rack being too wide
would not occur. This would then allow much lighter, simpler clevises to bolt to the end of the
rack and not apply any torsional moments about the long axis of the rack. Eliminating these
moments reduces the friction in the gear mesh and eradicates the problem of angular backlash in
the rack. This angular backlash was not discovered until the steering clevises, with their longer
lever arms to amplify the motion, were installed. If the rack had a slot with a linear bearing that
would prevent rotation of the rack this would not be an issue. Fortunately, the front wheels do
not transmit forward thrust and so the rack will take a seat in one orientation and stay there
once in motion. The angular backlash does cause toe changes and irritating steering slop at nearzero speed.
The front pushrod mounts to the top of the upright, and this placement was determined
initially by the need to clear the brake rotor but more importantly is the need to accommodate
large changes in camber while also allowing large steering angles up to 45 degrees (the heim
joints could not therefore be oriented as they are on the rear suspension). Ordinary heim joints
and ball joints did not have the right range of motion and/or were too bulky and heavy (although
a suspension designed only for a single, large negative camber angle could likely use a
traditional automotive ball joint). The solution was to use a clevis that holds the spherical bearing
on the end of the pushrod and which also rotates within the upright. This rotation is
accommodated by fiber-reinforced PTFE flange bushings from igus. The thrust face of the
bushing mates with a face lathed into the 303 stainless-steel clevis and there is a 3/8 shaft
protruding from that face and going through the bushings. This keeps the clevis aligned properly.
In the high camber configuration the pushrods force is almost directly in normal to the thrust
face of the bushing but the low camber configuration will put significant side-loads on the clevis
and the smooth rotation of the clevis is essential for steering. This clevis was turned on a lathe
from a large rod of stainless steel stock and then put into a mill to make the clevis mouth.
Because the machining was done accurately, the bushings have met and exceeded all
expectations of smooth rotation under combined loading. The bushings are pressed into a
machined tube that was welded into the top clevis of the front upright between the top control
arm pivot points. This tube adds rigidity to the upright and because of the small size of the
bushing it can be a small diameter. This therefore allows the top control arm pivots to be closer
together for a more consistent steering axis as the upright is steered. The pushrod and rocker
geometry is positioned so as to reduce the front end lift when the front wheels are steered. The
clevis is in between the two pivot points in the clevis in order to reduce the torque about the
steering axis caused by the force in the pushrod. The igus bushings are an essential component to
this design and were sourced from Rebel Racing Products. Figure 40 shows close-ups of the
clevis and the bushing area. A circlip and washer keep the clevis from coming out of the bushing
when all weight is off of the tire, such as if the car were driven off a jump or if it were on a jack.
The circlip only needs to support the weight of the upright and wheel and is unloaded in normal
operation.
56
Figure 40: Front Pushrod Rotating Clevis and Bushing Area
Figure 41 shows the high-camber front uprights and links assembled.
Figure 41: Front Uprights and Links
57
Figure 42 shows the installed rotating clevis piece for the front pushrod.
Figure 42: Installed Rotating Clevis
Figure 43 shows the lower control arm pivots and the steering rod attached to the front
upright. It also shows the precision, 4340 steel, high-angularity heim joint cone spacers which
were outsourced to a machinist with a CNC lathe because nearly 100 of these small but precision
components were required and the Formula Electric team did not have the time or skill to
manufacture them.
Figure 43: Front (High Camber) Lower Control Arm Pivots, Steering Pivot, and Heim Joint Spacers
Figure 44 shows all of the suspension corners laid-out together with the steering clevises.
58
Figure 44: Suspension Corners and Steering Clevises
The analysis of the suspension system was completed using free-body diagrams of each
upright. The forces at the tire contact patch were known from the cars expected mass properties
and the maximum longitudinal and lateral accelerations desired. With 6 two-force members, each
corner of the car is statically determinant and, using the known geometry, the forces in each link
can be found by solving the sum of forces and sum of moments equations simultaneously.
Simplifications to the geometry were made in hand calculations, which began with a front-view
free-body diagram and then used the front view forces as the resultants for each pseudo-A-arm
while satisfying the longitudinal force and moment equilibriums. A second, more accurate
analysis was performed (after the car was constructed) using a MATLAB program with inputs
being the coordinates of each pivot point obtained manually from inspection of a SolidWorks 3D
sketch for one steering angle at a time. The results from this study supported the hand
calculations and did not show any overloading. A more advanced analysis would incorporate the
use of vector loop equations (and the Newton-Raphson method due to the nonlinear
trigonometric functions involved) to find the exact geometry in MATLAB for any steering angle
or camber angle, then solving the load equations for each case and finding the maximum tensile
and compressive loads each link will encounter. To ensure the links are not extremely heavy or
bulky it must be acknowledged that the car cannot reach maximum lateral acceleration at the
same time as maximum longitudinal acceleration because of the friction circle effect. Therefore,
when the front suspension is steered the assumed braking force must be reduced while the lateral
force is increased and in a straight-ahead configuration the opposite is true. A reasonable factor
of safety can be ensured by maintaining the expected ratio of cornering to braking force but by
increasing each by the same amount. The front braking analysis must take into account the
outboard brakes, which require the links to react both a force and a torque on the upright since
59
the wheel bearing reacts the force from the caliper but the two are separated by the radius of the
brake rotor. The maximum deceleration assumed was 1.6g and the maximum lateral force was
assumed to be 2g with a car weight of 550 pounds (which turned out to be very close to the
actual weight) with a 180 pound driver.
It was discovered that the largest load during cornering at small steering angles was in the
pushrod, with very little in the lateral links. This is due to the angle of the pushrod in front view.
In the braking analysis and in the maximum-steering cornering simulation, forces through the
front, lower links were determined to be on the order of 2,000 pounds and less than 1,000 in the
rear suspension links. As a result, larger, heim joints would be necessary on the front bottom
links, although the #10 rod ends initially specified would suffice on the top links and in the rear.
To simplify the manufacturing, it was decided to use heim joints and bolts throughout the
suspension, a decision which was wise in hindsight considering the rough handling the links
received. The marginal weight gain from the larger rod ends was determined to be an acceptable
trade-off for simplicity and the extra safety margin. The spherical bearing rod ends are EM4
(right hand thread) and EML4 (left hand thread) from Rod End Supply, sourced through Rebel
Racing Products because they offered to sponsor the team by providing these rod ends at a much
reduced price.
6061-T6 aluminum alloy was chosen for the suspension links because the largest loads
are compressive and the extra yield strength of the more expensive 7075 alloy would not help
modulus-controlled buckling, since the elastic modulus of 7075 and 6061 is essentially the same.
An Euler buckling model was used, with an effective length of 14 inches, which is longer than
the greatest center to center span of all the links. A pin-pin connection was assumed since the
links have spherical bearings at each end. Thread pullout strength for 6061 was also determined
to have a factor of safety of about 2 after a test showed a tensile force of 3,000 pounds was
required to strip the internal threads out cleanly using a heim joint. A subsequent test using 7075
allow resulted in failure of the heim joints at the load expected based on manufacturer data. The
7075 internal threads showed no distortion and were still usable. The extra strength of the 7075
alloy is therefore primarily beneficial for maintaining thread quality over more fatigue cycles.
7075 was not in the original budget but some additional funds were procured in order to make
the more highly stressed front, lower, aft links from the stronger alloy. Rods of 0.5 inch diameter
were chosen for most of the links and but a diameter of 5/8 inch was necessary for the front
suspensions lower rear control arms which sees large enough compressive forces under braking
that the smaller diameter rods would buckle.
1/16 sheet steel was chosen for the outer walls of the uprights because it is easier to
weld to the 1/8 metal used for the clevises and the thick bearing race tube than thinner metal
(which would still have met strength requirements). It also has better dent resistance to keep the
uprights safe from unexpected impacts during construction and from cones or rocks on the race
track. The clevis pieces are thicker because they see out-of-plane loads. The front, top clevis
must react the pushrod force and this is done in the form of shear in the welds between the
bushing tube and the clevis pieces. The flat clevis pieces are therefore loaded in bending,
although the support span is short, being boxed in on three sides and with the tube itself adding
rigidity and linking the top and bottom flat clevis piece. The front, bottom clevis uses the thicker
metal because it is not boxed in as much because the lower links would hit any additional webs
60
of metal and restrict steering angle. The steering arm is two cantilevered pieces, but the majority
of the loads on them are in-plane where the section modulus is much greater. For the out-of plane
loads, the cantilever distance is short and the heim joint spacers add rigidity when the bolt
clamps them in. The thicker material adds minimal weight (since the clevis tabs are small) but
increases the rigidity of the steering system and makes the upright more durable. 4130
normalized chrome-molybdenum steel was chosen for a higher safety factor and because the
chassis is made from 4130 tubes so the team will be accustomed to welding it. The rear upright
was originally designed as a CNC-machined aluminum part but the large waste as well as the
unreliability of the availability of CNC machines in the required timeframe caused the decision
to move to a boxed, sheet steel weldment similar in construction to the front.
The uprights both were constructed using jigs (shown in Figure 45 and Figure 47)
consisting of steel plates with through-holes. Aluminum blocks bolted to the plate and then
upright pieces were bolted to the aluminum blocks and held in place for welding. The base of the
uprights were bearing race tubes made on a lathe and which had a lip on the inside diameter for
locating the FAG (front) and SKF (rear) tapered-roller wheel bearing races when they were
pressed into the tube until hitting the lip. A lip on the outside diameter mated with a sheet of steel
which formed one side of the main box of the upright. All the sheet metal for the uprights was
cut precisely on a water-jet machine and bent using a press-break. The SolidWorks sheet metal
tool was crucial for constructing the upright in CAD and then converting the pieces to flat
patterns with bend lines. Protrusions from the edges of the sheets mated with slots in other pieces
to locate the sheets relative to each other and to help hold them together for welding. The
uprights were self-jigging in this regard. This and the bearing cup lip are shown in Figures 45
and 46, depicting the high-camber front left upright.
The front, high-camber jig could be used for both left and right uprights without
modification while the rear jig could also be used for both left and right uprights but this required
relocating the clevis tab spacer blocks on the bottom. The blocks were taken off the jig plate and
used to make the top and bottom clevis sub-assemblies. These were made from a piece of square
4130 tube cut diagonally to make a right angle. This was used to box the clevises and make it
easier to weld up the rest of the main upright box. Care was needed to ensure the bottom clevises
were mirrored, as they are asymmetrical. The welding process involved putting one block with
two tabs on, welding both of those tabs, then inserting another block and tab and welding that
until a series of three clevises were welded on one end of the angle piece. Those were then bolted
to the jig plate along with the fourth clevis block and tabs. The outer tab for the fourth clevis was
welded and then the clevis blocks were removed from the jig plate so the inner tab of the fourth
clevis could be welded. Finally, the blocks were put back on the jig plate along with the simpler
top clevis piece and the rest of the sheet metal so that the whole upright could be welded. The
first piece to weld is around the bearing tube on the inside of the inner face sheet. Then the rest
can be done in any order. Revisions to the jig that would be made with the knowledge gleaned
from actually making an upright would include windows in the jig plate to allow some welding
of blind joints without taking the clevis pieces off the jig so much.
The front upright also began with making the top and bottom clevis pieces, then bolting
them to the jig along with the bearing tube and inner face sheet, welding the inner face sheet to
the bearing tube, and then welding the rest of the sides of the upright on.
61
Figure 45: Front Upright Jig and Partially-Constructed Left Front Upright
The low-camber front uprights require a different hole spacing in the jig plate as well as
different top and bottom clevis blocks. The same jig plate could have both hole patterns cut in it
so only one would be required. The center aluminum round piece which locates the bearing tube
can also be reused.
Figure 46: Tongue-and-Groove Joints for Self-Jigging
62
Figure 47: Rear Upright Jig
The suspension rockers are simple weldments consisting primarily of two triangular
sheets of 1/16 4130 steel sheetmetal and a tube connecting the two. The pivot points on the
chassis will have two inside diameter by 3/8 long igus L280 flange bushings pressed into a
0.95 wall thickness, outside diameter 4130 steel tube. These bushings are lighter, cheaper,
and smaller than ball bearings and do well when dirt gets in them. A 303 stainless-steel sleeve
will go around a bolt and have the diameter bearing surface. The stainless steel sleeve will
be clamped by the bolt between two 4130 steel tabs welded to the chassis. Stainless steel is one
of the best (lowest friction and wear) materials for use with igus L280.
The most challenging part of fabrication involved accommodating an inaccurate chassis
and a change in the chassis necessitated by an incomplete SolidWorks model of the drivetrain.
Jigs were fabricated to place the suspension tabs in the correct relative positions (particularly
important in the front for the steering geometry) and these helped very much when welding. The
simple 1/8 4130 steel tabs had to be re-made and notched in back to accommodate a complete
re-do of the rear half of the chassis, since a tube was interfering with the inner constant velocity
joint. This meant the suspension tabs had to rotate and become longer to compensate. The
rockers and springs also had to move but these are more flexible. In the front, one side is nearly
perfect because of the jig being used before the upper chassis tube was welded in, but the other
side has the top suspension point too high. Nothing could be done other than to move the
suspension point up as well because the angle of the clevis was important. Figure 48 is a
rendering of the front tab jig. It is made out of 6061-T6 aluminum alloy and can be used for both
63
right and left sides of the car by switching which holes the center piece connects to. The bottom
tabs clamp onto the square frame rail that makes up the floor of the chassis. The tops go over the
round tube that runs from the front roll hoop to the front bulkhead.
Figure 48: Front Suspension Tab Jig
Recommended Revisions
Reduce the kingpin inclination to avoid steering snatch.

Do not use drop-down clevises on the steering rack. Place the inner steering link
pivot point on the axis of the rack in order to reduce the stress on and friction in
the gear mesh. Obtain a steering rack that is narrow enough for the desired
kinematics or change the kinematics to suit the rack being used. Make the chassis
go around the rack if necessary, or else move the driver back so that the chassis
can end in front of his or her feet.
Increase the magnitude of the scrub radius to at least an inch positive or negative
to reduce the steering weight due to tire scrub at very slow speeds.
Use bearings instead of bushings for the rocker arms. The bushings require too
much precision on the running surface of the inner race. Bearings come with the
precision machining already done.
Put more extensive jigging in the budget so that the chassis can be made to fit the
suspension points and not the other way around. Link the left and right jigs of the
front suspension and rear suspension together so that each is kept in the same
plane. Include rocker arm pivot points and spring pivot points in the jigging,
along with steering rack mounts in the front. Linking the front and rear suspension
jigs together is not a high priority because the wheelbase can vary within the
current manufacturing tolerances without significant changes in the front-to-rear
weight distribution.
Make go-no-go gages/jigs for the links to make the suspension geometry more
accurate (though the pushrod and toe-links will likely need to be adjusted on the
car). These would be simple steel or aluminum plates with holes milled in them.
One rod end would be screwed in almost all the way and then bolted (with one
64
spacer cone) to the plate. The link would then be turned to adjust the center-tocenter length until a bolt could be put through the other rod end into the desired
hole in the jig (using another cone spacer to ensure the bolt is normal to the jig.
Use 7075 aluminum alloy and alloy steel rod ends for all links, if the budget
allows, in order to increase fatigue life and safety margins.
Design Verification
Structural Testing
Tests were carried out on the link assemblies using the model 1331 Instron Servo
Hydraulic Load Frame in the Cal Poly Composites Lab. Initial tests were done for thread pullout,
and more extensive tests done following a very unexpected failure of the right front suspension
on the maiden drive and which occurred during a relatively low-speed turn and without
impacting any obstacles. The driver was completely unhurt and the rest of the car was fine.
Link loads are known to a high degree of confidence because the 6 link system is
statically-determinant and the heim joints closely approximate a zero-moment joint (friction
means they can support a small moment and the threaded holes in the links cannot be perfectly
concentric with the rod, causing a moment). Therefore, overloading was not believed to be the
cause. Most heim joints were bent or broken and it was not known which joint broke first. One
probable cause is the improper handling of the links during the construction of the car. Against
the protestations of the suspension team, links were stepped on such that the aluminum rod bore
weight as a beam in bending. Links were also allowed to fall, sometimes with the weight of the
upright on them, and be caught by the misalignment cones, which produces a large bending
moment in the links, and, more importantly, the threaded section on each rod end. Additionally,
the threaded sections sometimes rested on the squared-off edge of the steel table while
supporting the weight of the car. In the future, much more care should be taken during assembly
and the suspension should be either assembled completely (on a corner-by corner basis) or taken
off entirely.
Due to the simple-to-manufacture design, a new link was made the day after the failure
(only one was damaged) and all heim joints were replaced on that corner, with the car running
briefly that night with the suspension functioning properly. Some of the rod ends from the failed
corner which did not appear to be damaged were tested the next day, along with several brand
new joints. At the end of the school year another test ended in a suspension failure, this time the
other front corner with the old heim joints. The driver reported steering snatch but it is not
known how much steering was applied when it happened, since the frame is not as wide as it
should have been in order to have the rack clevises act as steering stops. The car had made many
turns at moderate speeds before the failure, and since the side with new heim joints did not break
it is not known if the cause of both failures was the mishandled heim joints, too much steering
available, or errors in calculating link loads
Link buckling strength did not agree with Euler buckling calculations for a solid
aluminum rod with pinned ends and a length equal to the center-to-center distance between the
heim joint balls. The longer rod in Figure 49 showed a buckling strength of about 2250 pounds,
which is about what the Euler calculation is for an aluminum rod with pinned ends and a length
65
equal to only the aluminum part of the link, not the actual support span. The calculation for the
actual center-to-center distance gives a result of about 1750 pounds, close to where the short link
went nonlinear. The longer link assembly failed when the aluminum rod buckled in the first
mode shape.
2750
Compressive Load (pounds)
2500
2250
2000
1750
1500
12.5" center-to-center, 0.5"

diameter aluminum rod
1250
1000
13.25" center-to-center, 0.5"

diameter aluminum rod
750
500
250
0
0
0.02
0.04
0.06
0.08
0.1
Displacement (inches)
Figure 49: Link Buckling Test Results
The shorter link departed from a linear load-displacement curve at about 250 pounds
below the Euler buckling strength for the center-to-center distance. The failure mode was
bending of one of the heim joints at the threaded shank, just above the face of the aluminum rod
which it was screwed into. The exposed thread length was greater than that of the longer link, but
the heim joints were new. The rod ends on the long link were ones that appeared unbent but
which were from the failed right front corner. This suggests that there is a critical length for
exposed heim joint threads below which the aluminum rod fails and above which the heim joint
fails. In both cases the 6061-T6 aluminum threads were still usable after the test, supporting the
tensile testing in Figure 50 which shows that the pullout strength of these 0.5 inch diameter 6061
rods was about 3000 pounds. 7075 threads were also tested using a piece of 0.75 inch diameter
scrap but with the same -28 threads. The 7075 thread pullout strength could not be measured
because the heim joints would break first. The link was so strong that it was used for at least half
a dozen tensile and compression tests, even up to 4000 pounds, without damaging the interior
threads.
66
1/2" O.D. 6061-T6 Aluminum Rod,

1/4"-28 Interior Thread Pullout Test
3500
Tensile Load (pounds)
3000
2500
2000
1500
1000
500
0
0.000
0.010
0.020
0.030
0.040
Figure 50: 6061-T6 Aluminum Alloy Thread Pullout Test Data
The buckling of the rod ends themselves was then tested, with results shown in Figure 51.
It can be seen that exposing twice as much total shank length results in a lower effective modulus
for the link, because the series where both heim joints were extended shows a shallower slope to
the load-displacement curve. Also apparent is that the new heim joint took about 1500 pounds, or
about 66%, more load than the old heim joint which came from the failed front right corner but
which appeared undamaged. All of the rod ends in this test had performed either a tensile or link
buckling test beforehand but emerged seemingly undamaged. Any undetectable necking from
tension or bending from compression could cause early buckling. The softer release of load from
the old heim joints means they were probably slightly bent. The reason why the 0.649 inch
thread exposure was chosen is that this was as far into the 0.75 inch diameter 7075-T651 link
that one of the rod ends would go, even though the others would go further. This was one of the
joints in the double exposure test. A slight bend could cause the joint to not go all the way into
the rod. The new heim joint failed nearer to its theoretical compressive strength limit (3400
pounds) than its Euler buckling strength (24000 pounds), though the end condition for the length
of the shank was more complicated than pin-pin, having the spherical bearing housing at one end
and a rigidly fixed to a very stout aluminum link (about 6 inches long) at its other end with
another pin connection at the end of that. This buckling failure was very rapid and energetic,
releasing more load in a split second that the other joints held at their peaks. The nonlinearity of
the loading curve could be plastic deformation due to straight compressive loading rather than
bending. Figure 51 and Figure 50 show that predicting the buckling strength of a link is difficult.
67
4000
3750
3500
3250
Compressive Load (pounds)
3000
New Rod End,

Single 0.649"
Exposure
2750
2500
2250
Old Rod End,

Single 0.649"
Exposure
2000
1750
1500
Old Rod Ends,

Both 0.649"
Exposure
1250
1000
750
500
250
0
0
0.01
0.02
0.03
0.04
Figure 51: Rod End Shank Buckling Test
The tensile strength of the heim joints was twice what it was expected to be after
preliminary tests using similar looking rod ends found in a box of spare hardware from previous
Formula Electric cars. These broke at only 2300 pounds, or 700 pounds short of the 6061-T6
aluminum thread pullout strength. Fortunately, the heim joints obtained through Rebel Racing
Products met and exceeded their claimed ultimate strength of 3820 pounds. This can be seen in
Figure 52. The new heim joint was subjected to multiple pulls because the spherical bearings
began to slip out of the hydraulic grips on the first two runs. Clamp pressure was set low on the
first run in order to minimize the distortion out-of-round of the spherical bearing, which would
cause more friction in the rod end and change the stress state in the housing, possibly even
binding enough to change the end conditions on the links. For run 2 the clamp pressure was
raised but the grips still slipped. Run 3 had the spherical bearing inserted further into the clamps,
where the knurling was deeper, and this resulted in a good enough grip to break the joint on the
third run at a peak load of 3940 pounds. The joint behaved remarkably linearly (elastically) up to
3800 pounds, almost exactly its rated load. This means the joint has a lot of usable strength when
new, despite its low price and high angularity capability. The rod end from the failed suspension
corner went to a higher ultimate load of 4070 pounds, but departed from the linear region at only
2500 pounds, perhaps due to damage in the accident or due to slippage in the clamps since it
occurs near where the trend of the second run becomes nonlinear. The spherical bearing may
have been located in a slightly different location in the clamps which did not have as much grip.
Strain hardening due to accident damage would have resulted in a longer linear region and a
lower ultimate strength, which is likely what occurred for the new heim joint after three loading
68
cycles. The first three plots have been shifted so that they all begin at zero displacement, so any
hysteresis due to strain hardening cannot be seen. The rod end was not visibly deformed so the
plastic deformation was small and apparently the joint does not lose much ultimate strength from
the process. Therefore, peak loads greater than those due to the maneuvering forces of the
vehicle could produce some strain hardening, but the joint should still function normally and
have a greater elastic limit afterwards. There is about a 1000 pound margin before the onset of
strain hardening in the link with the highest load, which occurs in the front links during braking.
4500
4250
4000
3750
3500
Tensile Load (pounds)
3250
3000
2750
2500
New Rod End, Pull 1
2250
New Rod End, Pull 2
2000
New Rod End, Pull 3
1750
Old Rod End
1500
1250
1000
750
500
250
0
0
0.02
0.04
0.06
0.08
0.1
0.12
Figure 52: Tensile Strength of Heim Joints Sourced Through Rebel Racing Products
Testing Procedure
Research suggests that -40 degrees of camber is best for lateral acceleration but 0 degrees
is best for longitudinal acceleration. The intermediate goals of testing are to determine the
correlation between camber and lateral grip and the relationship between camber and
longitudinal grip. This data will then be used towards the final goals of developing a formula to
determine optimum camber angle for a particular track and determining if Polynx would have
produced more lateral acceleration.
The testing procedure is to first perform left and right skidpad tests with the high camber (-22
degrees) setup, measuring maximum lateral acceleration using a 3-axis accelerometer aligned to the
69
vehicle axes and mounted to the car at the location of the mass center when viewed from overhead.
Multiple laps will be driven in each direction and only the maximum sustained lateral acceleration will be
used to determine performance. The lateral acceleration will be corrected for the roll of the car by using
the data from the vertical channel, which should be 1g if the car is flat. After skidpad tests, maximum
thrust will be determined with the same accelerometer via straight-line acceleration tests.
After the high-camber tests the right side suspension corners will be switched to the 0 camber
configuration. This will simulate the Polynx dynamic camber design. A right skidpad test will then be
performed to determine how much, if any, the inside tire is hampered by the negative camber.
Next, the left suspension will be reconfigured to low camber and the same skidpad and
longitudinal acceleration tests will be performed as were with the high camber configuration. If available,
square-shouldered tires would be used at this point as another comparison. The longitudinal grip is
expected to increase but the lateral acceleration to decrease.
To cancel-out any asymmetries in the chassis and suspension points, the right suspension will
then be changed back to a high camber setup and a left skidpad test performed. Following that, the left
side will be reset to high camber as well so that the car is symmetric and can be driven for purposes other
than camber tire testing.
After careful reading of the existing data on camber cars, it appears that there should not be a
significant gain in lateral acceleration (on the order of 0.05g) when the inside tire is vertical as opposed to
negatively-cambered. If there is, however, then Polynx is justified. The rest of the data will be plotted and
curves fit to them so that the relations between camber and lateral grip and camber and longitudinal grip
can be determined. To develop a formula for the optimal negative camber angle it must be known how
longitudinal acceleration and lateral acceleration improve lap time at a particular circuit. A first attempt at
this would involve going from steady-state turning through the arc length of a turn to full acceleration to
full braking (the braking point would need to be back-calculated knowing the entry speed to the corner
and assuming a constant deceleration) to steady-state turning, etc. and therefore ignoring transients. This
would give a good idea of how important lateral grip and longitudinal grip are for lap time and the
relations found with the test data would then be inserted into the lap time formula, leaving one variable of
camber. Minimizing the lap time function would then yield the optimal camber angle. The relation
between camber and lateral grip and the correlation between camber and longitudinal grip could be
approximated as linear with just the two points of data, but the turnbuckle adjustment of the suspension
links allows for several camber angles near each of the extremes to be examined as well. This would give
information on how nonlinear the relationships are. A linear approximation may or may not be sufficient.
Further skidpad testing would involve shifting ballast to change the front-to-rear weight
distribution and to change the mass center height. Additional tests would include a steering
wheel position sensor in combination with the accelerometer so that transient response can be
measured along with the oversteer/understeer balance. Linear potentiometers attached to the
suspension rocker (with a known motion ratio) could be used to tune dampers and to study the
lateral force response to weight transfer with negatively-cambered tires (effectively a relaxation
length). The transient test results could then be used to refine the torque vectoring control
system.
70
References
Milliken, William F. Equations of Motion. Cambridge MA, U.S.A.: Bentley Publishers, 2006.
Print.
Milliken, William F., & Douglas L. Milliken. Race Car Vehicle Dynamics. Warrendale, PA,
U.S.A.: SAE International, 1995. Print.
Pacejka, Hans B. Tire and Vehicle Dynamics. 3rd ed. Oxford, UK: Elsevier Ltd., 2012. Print.
Cossalter, Vittore. Motorcycle Dynamics. 2nd English edition. Lexington, KY, U.S.A.: Lulu,
2006. Print.
Fonda, Albert G. Tyre Tests and Interpretation of Data. Sage Publications on behalf of the
Institution for Mechanical Engineers, 1956. PDF.
Formula Electric SAE Advisor: Professor Fabijanic, jfabijan@calpoly.edu, 805-756-5547
71
Appendix A: Simulink Modeling
72
Appendix B: Vendors
Aircraft Spruce
http://www.aircraftspruce.com/
Speedy Metals
http://www.speedymetals.com/
McMaster Carr
http://www.mcmaster.com/#
Online Metals
http://www.onlinemetals.com/
Rebel Racing Products
http://rebelracingproducts.com/
Rod End Supply
http://www.rodendsupply.com/index.php
Grainger
http://www.grainger.com/Grainger/wwg/start.shtml
Mid-South Minimoto
http://www.midsouthminimoto.com/
SKF Bearings
http://www.skf.com/group/products/bearings-units-housings/index.html
FAG Bearings
http://www.fag.com/content.fag.de/en/index.jsp
igus
http://www.igus.com/default.asp?c=us&L=en
73

Camber-Car Design & Dynamics Document Optimized 40 Character Title

Enviado por

Dados do documento

Descrição original:

Título original

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

Camber-Car Design & Dynamics Document Optimized 40 Character Title

Enviado por

Direitos autorais:

Formatos disponíveis

Camber-Car Design & Dynamics

Tristan French, E.I.T.

Mechanical Engineering Department

Figure 25: Steering Mechanism Shown on a Simplified (2 Pivot) Polynx Suspension................ 40

Abstract & Background

Fundamental Tire & Vehicle Dynamics

Cossalter, Vittore: Motorcycle Dynamics, 2nd ed., page 56

Figure 1: SAE Tire Axes

Cossalter, Vittore: Motorcycle Dynamics, 2nd ed., page 56

Figure 3: Typical Load Sensitivity Behavior

Figure 4: Shape of Deformed Contact Patch for Frictionless Road

Cossalter, Vittore: Motorcycle Dynamics, 2nd ed., page 67

Combining Inclination & Slip Angle

Cossalter, Vittore: Motorcycle Dynamics, 2nd ed., page 49

Figure 8: Milliken MX-1 Front View

Figure 9: Milliken MX-1 Side View

Photo from ultimatecarspage.com via Google Images

Milliken, William F.: Equations of Motion, page 521

Milliken, William F.: Equations of Motion, page 522

Round-Section Tires vs. Square-Shouldered Tires

Figure 15: Peak Lateral Force vs. Inclination Angle

Figure 16: Tires Used to Generate Data for Figure 9

Figure 17: Lateral Force Characteristics for Various Round-Shouldered Tires

Figure 18: Friction Circle from Sennas Lotus at 1987 Australian GP

Figure 19: Ackermann Steering Geometry

Figure 20: Steering Geometry Definitions

Milliken, William F.: Equations of Motion, page 524,525

Cossalter, Vittore.: Motorcycle Dynamics, pages 22,23

Figure 21: Instant Center vs. True Center

Figure 22: Roll Center Construction

Figure 23: Vertical Jacking Force Due to Applied Lateral Force

Milliken, William F.: Equations of Motion, page 517

Cossalter, Vittore: Motorcycle Dynamics, 2nd ed., page 56

Milliken, William F.: Equations of Motion, page 520

using the formula a y,max

, where ay,max is the maximum lateral acceleration in g, hcg is

Fonda, Albert G.: Tire Tests and Interpretation of Data,

Milliken, William F.: Equations of Motion, page 517

Figure 24: General Polynx Design

Figure 25: Steering Mechanism Shown on a Simplified (2 Pivot) Polynx Suspension

Final Design: Static Camber

Figure 26: Front Suspension Pivot Point Front View Geometry

Figure 27: Rear Suspension Pivot Point Front View Geometry

Figure 28: Side View of Front Suspension Points

Figure 29: Side View of Rear Suspension Points

Figure 30: Front View of Front Upright Geometry

Figure 31: Front View of Rear Upright Geometry

Figure 33: Top View of Left Front, High Camber Upright

Figure 34: Side View of Rear Upright Geometry

Table 1: Link Lengths

Center-to-Center Length (inches)

Inner Wheel Angle

Steering Wheel Angle (deg)

Bump Travel (inches)

Milliken, William F.: Equations of Motion, page 515

Figure 37: Final Suspension Design

Figure 38: Right Rear Suspension Corner

Figure 39: Anti-Bumpsteer Steering Rack Clevises

Figure 40: Front Pushrod Rotating Clevis and Bushing Area

Figure 41 shows the high-camber front uprights and links assembled.

Figure 41: Front Uprights and Links

Figure 42: Installed Rotating Clevis